Merge pull request #3984 from facebook/dev

v1.5.6

.circleci/images/primary/Dockerfile

@@ -1,4 +1,4 @@
-FROM circleci/buildpack-deps:focal
+FROM circleci/buildpack-deps@sha256:f6f10c11b7b8ccfd4f4a5b830c3256803604ce61292b60cb22e26b12f62b0e8c
 
 RUN sudo dpkg --add-architecture i386
 RUN sudo apt-get -y -qq update

.cirrus.yml

@@ -2,8 +2,8 @@ task:
   name: FreeBSD (shortest)
   freebsd_instance:
     matrix:
-      image_family: freebsd-13-0
-      image_family: freebsd-12-2
+      image_family: freebsd-14-0
+      image_family: freebsd-13-2
   install_script: pkg install -y gmake coreutils
   script: |
     MOREFLAGS="-Werror" gmake -j all

.github/workflows/commit.yml

@@ -0,0 +1,89 @@
+name: facebook/zstd/commit
+on:
+  push:
+    branches:
+    - dev
+permissions: read-all
+jobs:
+  short-tests-0:
+    runs-on: ubuntu-latest
+    services:
+      docker:
+        image: fbopensource/zstd-circleci-primary:0.0.1
+        options: --entrypoint /bin/bash
+    steps:
+    - uses: actions/checkout@v4
+    - name: Install Dependencies
+      run: |
+        sudo apt-get update
+        sudo apt-get install libcurl4-gnutls-dev
+    - name: Test
+      run: |
+        ./tests/test-license.py
+        cc -v
+        CFLAGS="-O0 -Werror -pedantic" make allmost; make clean
+        make c99build; make clean
+        make c11build; make clean
+        make -j regressiontest; make clean
+        make shortest; make clean
+        make cxxtest; make clean
+  short-tests-1:
+    runs-on: ubuntu-latest
+    services:
+      docker:
+        image: fbopensource/zstd-circleci-primary:0.0.1
+        options: --entrypoint /bin/bash
+    steps:
+    - uses: actions/checkout@v4
+    - name: Install Dependencies
+      run: |
+        sudo apt-get update
+        sudo apt-get install gcc-powerpc-linux-gnu gcc-arm-linux-gnueabi gcc-aarch64-linux-gnu libc6-dev-ppc64-powerpc-cross libcurl4-gnutls-dev lib64gcc-11-dev-powerpc-cross
+    - name: Test
+      run: |-
+        make gnu90build; make clean
+        make gnu99build; make clean
+        make ppc64build V=1; make clean
+        make ppcbuild   V=1; make clean
+        make armbuild   V=1; make clean
+        make aarch64build V=1; make clean
+        make -C tests test-legacy test-longmatch; make clean
+        make -C lib libzstd-nomt; make clean
+  regression-test:
+    runs-on: ubuntu-latest
+    services:
+      docker:
+        image: fbopensource/zstd-circleci-primary:0.0.1
+        options: --entrypoint /bin/bash
+    env:
+      CIRCLE_ARTIFACTS: "/tmp/circleci-artifacts"
+    steps:
+    - uses: actions/checkout@v4
+    - name: restore_cache
+      uses: actions/cache@v4
+      with:
+        key: regression-cache-{{ checksum "tests/regression/data.c" }}-v0
+        path: tests/regression/cache
+        restore-keys: regression-cache-{{ checksum "tests/regression/data.c" }}-v0
+    - name: Install Dependencies
+      run: |
+        sudo apt-get update
+        sudo apt-get install libcurl4-gnutls-dev
+    - name: Regression Test
+      run: |
+        make -C programs zstd
+        make -C tests/regression test
+        mkdir -p $CIRCLE_ARTIFACTS
+        ./tests/regression/test                     \
+            --cache  tests/regression/cache         \
+            --output $CIRCLE_ARTIFACTS/results.csv  \
+            --zstd   programs/zstd
+        echo "NOTE: The new results.csv is uploaded as an artifact to this job"
+        echo "      If this fails, go to the Artifacts pane in CircleCI, "
+        echo "      download /tmp/circleci-artifacts/results.csv, and if they "
+        echo "      are still good, copy it into the repo and commit it."
+        echo "> diff tests/regression/results.csv $CIRCLE_ARTIFACTS/results.csv"
+        diff tests/regression/results.csv $CIRCLE_ARTIFACTS/results.csv
+    - uses: actions/upload-artifact@v4
+      with:
+        path: "/tmp/circleci-artifacts"

.github/workflows/dev-long-tests.yml

@@ -9,11 +9,13 @@ on:
   pull_request:
     branches: [ dev, release, actionsTest ]
 
+permissions: read-all
+
 jobs:
   make-all:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: make all
       run: make all
 
@@ -24,7 +26,7 @@ jobs:
       DEVNULLRIGHTS: 1
       READFROMBLOCKDEVICE: 1
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: make test
       run: make test
 
@@ -32,36 +34,50 @@ jobs:
   make-test-osx:
     runs-on: macos-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: OS-X test
       run: make test # make -c lib all doesn't work because of the fact that it's not a tty
 
+  # lasts ~24mn
+  make-test-32bit:
+    runs-on: ubuntu-latest
+    env:
+      DEVNULLRIGHTS: 1
+      READFROMBLOCKDEVICE: 1
+    steps:
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
+    - name: make test
+      run: |
+        sudo apt-get -qqq update
+        make libc6install
+        CFLAGS="-m32" make test
+
   no-intrinsics-fuzztest:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: no intrinsics fuzztest
       run: MOREFLAGS="-DZSTD_NO_INTRINSICS" make -C tests fuzztest
 
   tsan-zstreamtest:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-20.04
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: thread sanitizer zstreamtest
       run: CC=clang ZSTREAM_TESTTIME=-T3mn make tsan-test-zstream
 
   ubsan-zstreamtest:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-20.04
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: undefined behavior sanitizer zstreamtest
       run: CC=clang make uasan-test-zstream
 
   # lasts ~15mn
   tsan-fuzztest:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-20.04
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: thread sanitizer fuzztest
       run: CC=clang make tsan-fuzztest
 
@@ -69,7 +85,7 @@ jobs:
   big-tests-zstreamtest32:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: zstream tests in 32bit mode, with big tests
       run: |
         sudo apt-get -qqq update
@@ -78,9 +94,9 @@ jobs:
 
   # lasts ~23mn
   gcc-8-asan-ubsan-testzstd:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-20.04
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: gcc-8 + ASan + UBSan + Test Zstd
       # See https://askubuntu.com/a/1428822
       run: |
@@ -90,16 +106,16 @@ jobs:
         CC=gcc-8 make -j uasan-test-zstd </dev/null V=1
 
   clang-asan-ubsan-testzstd:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-20.04
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: clang + ASan + UBSan + Test Zstd
       run: CC=clang make -j uasan-test-zstd </dev/null V=1
 
   gcc-asan-ubsan-testzstd-32bit:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-20.04
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: ASan + UBSan + Test Zstd, 32bit mode
       run: |
         sudo apt-get -qqq update
@@ -111,9 +127,9 @@ jobs:
     # so any data coming from these libraries is always considered "uninitialized"
 
   gcc-8-asan-ubsan-fuzz:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-20.04
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: gcc-8 + ASan + UBSan + Fuzz Test
       # See https://askubuntu.com/a/1428822
       run: |
@@ -123,16 +139,16 @@ jobs:
         CC=gcc-8 FUZZER_FLAGS="--long-tests" make clean uasan-fuzztest
 
   clang-asan-ubsan-fuzz:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-20.04
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: clang + ASan + UBSan + Fuzz Test
       run: CC=clang FUZZER_FLAGS="--long-tests" make clean uasan-fuzztest
 
   gcc-asan-ubsan-fuzz32:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-20.04
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: ASan + UBSan + Fuzz Test 32bit
       run: |
         sudo apt-get -qqq update
@@ -140,9 +156,9 @@ jobs:
         CFLAGS="-O3 -m32" FUZZER_FLAGS="--long-tests" make uasan-fuzztest
 
   clang-asan-ubsan-fuzz32:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-20.04
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: clang + ASan + UBSan + Fuzz Test 32bit
       run: |
         sudo apt-get -qqq update
@@ -150,30 +166,40 @@ jobs:
         CC=clang CFLAGS="-O3 -m32" FUZZER_FLAGS="--long-tests" make uasan-fuzztest
 
   asan-ubsan-regression:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-20.04
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: ASan + UBSan + Regression Test
       run: make -j uasanregressiontest
 
   clang-ubsan-regression:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-20.04
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: clang + ASan + UBSan + Regression Test
       run: CC=clang make -j uasanregressiontest
 
   msan-regression:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-20.04
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: MSan + Regression Test
       run: make -j msanregressiontest
 
-  clang-msan-fuzz:
-    runs-on: ubuntu-latest
+  clang-msan-fuzz-unoptimized:
+    runs-on: ubuntu-20.04
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+      - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
+      - name: clang + MSan + Fuzz Test
+        run: |
+          sudo apt-get -qqq update
+          sudo apt-get install clang
+          CC=clang MOREFLAGS="-O0" make clean msan-fuzztest
+
+  clang-msan-fuzz:
+    runs-on: ubuntu-20.04
+    steps:
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: clang + MSan + Fuzz Test
       run: |
         sudo apt-get -qqq update
@@ -182,9 +208,9 @@ jobs:
 
   # lasts ~24mn
   clang-msan-testzstd:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-20.04
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: clang + MSan + Test Zstd
       run: |
         sudo apt-get update
@@ -194,7 +220,7 @@ jobs:
   armfuzz:
       runs-on: ubuntu-latest
       steps:
-      - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+      - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
       - name: Qemu ARM emulation + Fuzz Test
         run: |
           sudo apt-get -qqq update
@@ -204,7 +230,7 @@ jobs:
   valgrind-fuzz-test:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: valgrind + fuzz test stack mode    # ~ 7mn
       shell: 'script -q -e -c "bash {0}"'
       run: |
@@ -220,8 +246,8 @@ jobs:
       run:
         shell: msys2 {0}
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
-    - uses: msys2/setup-msys2@v2
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
+    - uses: msys2/setup-msys2@cc11e9188b693c2b100158c3322424c4cc1dadea # tag=v2.22.0
       with:
         msystem: MINGW64
         install: make
@@ -243,7 +269,7 @@ jobs:
 
   # lasts ~20mn
   oss-fuzz:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-20.04
     strategy:
       fail-fast: false
       matrix:
@@ -264,7 +290,7 @@ jobs:
         dry-run: false
         sanitizer: ${{ matrix.sanitizer }}
     - name: Upload Crash
-      uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce # tag=v3.1.2
+      uses: actions/upload-artifact@5d5d22a31266ced268874388b861e4b58bb5c2f3 # tag=v4.3.1
       if: failure() && steps.build.outcome == 'success'
       with:
         name: ${{ matrix.sanitizer }}-artifacts

.github/workflows/dev-short-tests.yml

@@ -10,25 +10,27 @@ on:
   pull_request:
     branches: [ dev, release, actionsTest ]
 
+permissions: read-all
+
 jobs:
   linux-kernel:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: linux kernel, library + build + test
       run: make -C contrib/linux-kernel test CFLAGS="-Werror -Wunused-const-variable -Wunused-but-set-variable"
 
   benchmarking:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: make benchmarking
       run: make benchmarking
 
   check-32bit: # designed to catch https://github.com/facebook/zstd/issues/2428
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: make check on 32-bit
       run: |
         sudo apt update
@@ -36,9 +38,9 @@ jobs:
         CFLAGS="-m32 -O1 -fstack-protector" make check V=1
 
   check-x32:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-20.04  # ubuntu-latest == ubuntu-22.04 have issues currently with x32
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: make check on x32 ABI # https://en.wikipedia.org/wiki/X32_ABI
       env:
         CHECK_CONSTRAINED_MEM: true
@@ -50,16 +52,24 @@ jobs:
   build-c89:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: ensure zstd can be build with c89/c90 compilers (+ long long support + variadic macros)
       run: |
         make c89build V=1
 
+  build-zstd-dll:
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
+    - name: build zstd bin against a dynamic lib (debuglevel for more dependencies)
+      run: |
+        make -C lib lib-mt-release
+        DEBUGLEVEL=2 make -C programs zstd-dll
 
   gcc-7-libzstd:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: gcc-7 + libzstdmt compilation
       # See https://askubuntu.com/a/1428822
       run: |
@@ -70,24 +80,21 @@ jobs:
         make clean
         LDFLAGS=-Wl,--no-undefined make -C lib libzstd-mt
 
-    # candidate test (to check) : underlink test
+    # candidate test (for discussion) : underlink test
     # LDFLAGS=-Wl,--no-undefined : will make the linker fail if dll is underlinked
 
   cmake-build-and-test-check:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
-    - name: cmake build and test check
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
+    - name: cmake build and test
       run: |
-        FUZZERTEST=-T1mn ZSTREAM_TESTTIME=-T1mn make cmakebuild
-        cp -r ./ "../zstd source"
-        cd "../zstd source"
-        FUZZERTEST=-T1mn ZSTREAM_TESTTIME=-T1mn make cmakebuild
+        FUZZERTEST=-T1mn ZSTREAM_TESTTIME=-T1mn make cmakebuild V=1
 
   cpp-gnu90-c99-compatibility:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: C++, gnu90 and c99 compatibility
       run: |
         make cxxtest
@@ -101,7 +108,7 @@ jobs:
   mingw-cross-compilation:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: mingw cross-compilation
       run: |
         # sudo update-alternatives --set x86_64-w64-mingw32-g++ /usr/bin/x86_64-w64-mingw32-g++-posix; (doesn't work)
@@ -112,7 +119,7 @@ jobs:
   armbuild:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: ARM Build Test
       run: |
         sudo apt-get -qqq update
@@ -122,7 +129,7 @@ jobs:
   bourne-shell:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: Bourne shell compatibility (shellcheck)
       run: |
         wget https://github.com/koalaman/shellcheck/releases/download/v0.7.1/shellcheck-v0.7.1.linux.x86_64.tar.xz
@@ -132,7 +139,7 @@ jobs:
   zlib-wrapper:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: zlib wrapper test
       run: |
         sudo apt-get -qqq update
@@ -143,7 +150,7 @@ jobs:
   lz4-threadpool-libs:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: LZ4, thread pool, and libs build testslib wrapper test
       run: |
         make lz4install
@@ -157,7 +164,7 @@ jobs:
   gcc-make-tests-32bit:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: Make all, 32bit mode
       run: |
         sudo apt-get -qqq update
@@ -167,7 +174,7 @@ jobs:
   gcc-8-make:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+      - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
       - name: gcc-8 build
         # See https://askubuntu.com/a/1428822
         run: |
@@ -176,10 +183,29 @@ jobs:
           make gcc8install
           CC=gcc-8 CFLAGS="-Werror" make -j all
 
+  make-external-compressors:
+    strategy:
+      matrix:
+        include:
+          - name: "no external compressors"
+            flags: "HAVE_ZLIB=0 HAVE_LZ4=0 HAVE_LZMA=0"
+          - name: "only zlib"
+            flags: "HAVE_ZLIB=1 HAVE_LZ4=0 HAVE_LZMA=0"
+          - name: "only lz4"
+            flags: "HAVE_ZLIB=0 HAVE_LZ4=1 HAVE_LZMA=0"
+          - name: "only lzma"
+            flags: "HAVE_ZLIB=0 HAVE_LZ4=0 HAVE_LZMA=1"
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
+      - name: Build with ${{matrix.name}}
+        run: ${{matrix.flags}} make zstd
+
+
   implicit-fall-through:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+      - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
       - name: -Wimplicit-fallthrough build
         run: |
           make clean
@@ -190,7 +216,7 @@ jobs:
   meson-linux:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+      - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
       - name: Install packages
         run: |
           sudo apt-get update
@@ -214,17 +240,15 @@ jobs:
   meson-windows:
     runs-on: windows-latest
     steps:
-      - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+      - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
       - name: Install packages
         run: pip install --pre meson
-      - name: Initialize the MSVC dev command prompt
-        uses: ilammy/msvc-dev-cmd@cec98b9d092141f74527d0afa6feb2af698cfe89
       - name: Configure with Meson
         run: |
-          meson setup build/meson/ builddir -Dbin_tests=true -Dbin_programs=true -Dbin_contrib=true
+          meson setup --vsenv build/meson/ builddir -Dbin_tests=true -Dbin_programs=true -Dbin_contrib=true
       - name: Build with Meson
         run: |
-          ninja -C builddir/
+          meson compile -C builddir/
       - name: Test with Meson
         run: |
           meson test -C builddir/ --print-errorlogs
@@ -241,50 +265,64 @@ jobs:
           - generator: "Visual Studio 17 2022"
             flags: "-A Win32"
           - generator: "MinGW Makefiles"
+          - generator: "Visual Studio 17 2022"
+            flags: "-T ClangCL"
     runs-on: windows-2022
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: Add MSBuild to PATH
-      uses: microsoft/setup-msbuild@v1.3
-    - name: Build
+      uses: microsoft/setup-msbuild@6fb02220983dee41ce7ae257b6f4d8f9bf5ed4ce # tag=v2.0.0
+    - name: Build & Test
       working-directory: ${{env.GITHUB_WORKSPACE}}
       run: |
         cd build\cmake
         mkdir build
         cd build
-        cmake.exe -G "${{matrix.generator}}" ${{matrix.flags}} ..
+        cmake.exe -G "${{matrix.generator}}" ${{matrix.flags}} -DCMAKE_BUILD_TYPE=Debug -DZSTD_BUILD_TESTS:BOOL=ON -DZSTD_ZSTREAM_FLAGS=-T30s -DZSTD_FUZZER_FLAGS=-T30s -DZSTD_FULLBENCH_FLAGS=-i0 ..
         cmake.exe --build .
+        ctest.exe -V -C Debug
 
   msbuild-visual-studio:
     strategy:
+      fail-fast: false  # 'false' means Don't stop matrix workflows even if some matrix failed.
       matrix:
         include: [
-          { name: "VS 2022 x64 Debug", platform: x64, configuration: Debug, toolset: v143, runner: "windows-2022"},
-          { name: "VS 2022 Win32 Debug", platform: Win32, configuration: Debug, toolset: v143, runner: "windows-2022"},
-          { name: "VS 2022 x64 Release", platform: x64, configuration: Release, toolset: v143, runner: "windows-2022"},
-          { name: "VS 2022 Win32 Release", platform: Win32, configuration: Release, toolset: v143, runner: "windows-2022"},
-          { name: "VS 2019 x64 Release", platform: Win32, configuration: Release, toolset: v142, runner: "windows-2019"},
-          { name: "VS 2019 Win32 Release", platform: x64, configuration: Release, toolset: v142, runner: "windows-2019"},
+          { name: "VS 2022 x64 Debug", platform: x64, configuration: Debug, toolset: v143, runner: "windows-2022", arch: "" },
+          { name: "VS 2022 Win32 Debug", platform: Win32, configuration: Debug, toolset: v143, runner: "windows-2022", arch: "" },
+          { name: "VS 2022 x64 Release", platform: x64, configuration: Release, toolset: v143, runner: "windows-2022", arch: ""},
+          { name: "VS 2022 Win32 Release", platform: Win32, configuration: Release, toolset: v143, runner: "windows-2022", arch: ""},
+          { name: "VS 2019 x64 Release", platform: Win32, configuration: Release, toolset: v142, runner: "windows-2019", arch: ""},
+          { name: "VS 2019 Win32 Release", platform: x64, configuration: Release, toolset: v142, runner: "windows-2019", arch: ""},
+          { name: "VS 2022 x64 Release AVX2", platform: x64, configuration: Release, toolset: v143, runner: "windows-2022", arch: "AdvancedVectorExtensions2" },
         ]
     runs-on: ${{matrix.runner}}
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: Add MSBuild to PATH
-      uses: microsoft/setup-msbuild@v1.3
-    - name: Build
+      uses: microsoft/setup-msbuild@6fb02220983dee41ce7ae257b6f4d8f9bf5ed4ce # tag=v2.0.0
+    - name: Build ${{matrix.name}}
       working-directory: ${{env.GITHUB_WORKSPACE}}
       # See https://docs.microsoft.com/visualstudio/msbuild/msbuild-command-line-reference
+      if: ${{ matrix.arch == '' }}
       run: >
         msbuild "build\VS2010\zstd.sln" /m /verbosity:minimal /property:PlatformToolset=${{matrix.toolset}}
-        /t:Clean,Build /p:Platform=${{matrix.platform}} /p:Configuration=${{matrix.configuration}}
+        /t:Clean,Build /p:Platform=${{matrix.platform}} /p:Configuration=${{matrix.configuration}} /warnaserror
+    - name: Build ${{matrix.name}}
+      working-directory: ${{env.GITHUB_WORKSPACE}}
+      # See https://docs.microsoft.com/visualstudio/msbuild/msbuild-command-line-reference
+      if: ${{ matrix.arch != '' }}
+      run: >
+        msbuild "build\VS2010\zstd.sln" /m /verbosity:minimal /property:PlatformToolset=${{matrix.toolset}}
+        /t:Clean,Build /p:Platform=${{matrix.platform}} /p:Configuration=${{matrix.configuration}} /warnaserror
+        /p:InstructionSet=${{matrix.arch}}
 
-  # This tests that we don't accidently grow the size too much.
+  # This tests that we don't accidentally grow the size too much.
   # If the size grows intentionally, you can raise these numbers.
   # But we do need to think about binary size, since it is a concern.
   libzstd-size:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: libzstd size test
       run: |
         make clean && make -j -C lib libzstd && ./tests/check_size.py lib/libzstd.so 1100000
@@ -295,7 +333,7 @@ jobs:
   minimal-decompressor-macros:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: minimal decompressor macros
       run: |
         make clean && make -j all ZSTD_LIB_MINIFY=1 MOREFLAGS="-Werror"
@@ -306,11 +344,13 @@ jobs:
         make clean && make check MOREFLAGS="-Werror -DHUF_FORCE_DECOMPRESS_X2 -DZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG"
         make clean && make -j all MOREFLAGS="-Werror -DZSTD_NO_INLINE -DZSTD_STRIP_ERROR_STRINGS"
         make clean && make check MOREFLAGS="-Werror -DZSTD_NO_INLINE -DZSTD_STRIP_ERROR_STRINGS"
+        make clean && make check ZSTD_LIB_EXCLUDE_COMPRESSORS_DFAST_AND_UP=1 MOREFLAGS="-Werror"
+        make clean && make check ZSTD_LIB_EXCLUDE_COMPRESSORS_GREEDY_AND_UP=1 MOREFLAGS="-Werror"
 
   dynamic-bmi2:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: dynamic bmi2 tests
       run: |
         make clean && make -j check MOREFLAGS="-O0 -Werror -mbmi2"
@@ -322,13 +362,12 @@ jobs:
   test-variants:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: make all variants & validate
       run: |
         make -j -C programs allVariants MOREFLAGS=-O0
         ./tests/test-variants.sh
 
-
   qemu-consistency:
     name: QEMU ${{ matrix.name }}
     runs-on: ubuntu-20.04
@@ -342,13 +381,15 @@ jobs:
           { name: PPC64LE,  xcc_pkg: gcc-powerpc64le-linux-gnu, xcc: powerpc64le-linux-gnu-gcc, xemu_pkg: qemu-system-ppc,    xemu: qemu-ppc64le-static },
           { name: S390X,    xcc_pkg: gcc-s390x-linux-gnu,       xcc: s390x-linux-gnu-gcc,       xemu_pkg: qemu-system-s390x,  xemu: qemu-s390x-static   },
           { name: MIPS,     xcc_pkg: gcc-mips-linux-gnu,        xcc: mips-linux-gnu-gcc,        xemu_pkg: qemu-system-mips,   xemu: qemu-mips-static    },
+          { name: RISC-V,   xcc_pkg: gcc-riscv64-linux-gnu,     xcc: riscv64-linux-gnu-gcc,     xemu_pkg: qemu-system-riscv64,xemu: qemu-riscv64-static },
           { name: M68K,     xcc_pkg: gcc-m68k-linux-gnu,        xcc: m68k-linux-gnu-gcc,        xemu_pkg: qemu-system-m68k,   xemu: qemu-m68k-static    },
+          { name: SPARC,    xcc_pkg: gcc-sparc64-linux-gnu,     xcc: sparc64-linux-gnu-gcc,     xemu_pkg: qemu-system-sparc,  xemu: qemu-sparc64-static },
         ]
     env:                        # Set environment variables
       XCC: ${{ matrix.xcc }}
       XEMU: ${{ matrix.xemu }}
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: apt update & install
       run: |
         sudo apt-get update
@@ -368,6 +409,7 @@ jobs:
     - name: ARM64
       if: ${{ matrix.name == 'ARM64' }}
       run: |
+        LDFLAGS="-static -z force-bti" MOREFLAGS="-mbranch-protection=standard" CC=$XCC QEMU_SYS=$XEMU make clean check
         LDFLAGS="-static" CC=$XCC QEMU_SYS=$XEMU make clean check
     - name: PPC
       if: ${{ matrix.name == 'PPC' }}
@@ -385,10 +427,18 @@ jobs:
       if: ${{ matrix.name == 'MIPS' }}
       run: |
         LDFLAGS="-static" CC=$XCC QEMU_SYS=$XEMU make clean check
+    - name: RISC-V
+      if: ${{ matrix.name == 'RISC-V' }}
+      run: |
+        LDFLAGS="-static" CC=$XCC QEMU_SYS=$XEMU make clean check
     - name: M68K
       if: ${{ matrix.name == 'M68K' }}
       run: |
         LDFLAGS="-static" CC=$XCC QEMU_SYS=$XEMU make clean check
+    - name: SPARC
+      if: ${{ matrix.name == 'SPARC' }}
+      run: |
+        LDFLAGS="-static" CC=$XCC QEMU_SYS=$XEMU make clean check
 
   mingw-short-test:
     runs-on: windows-latest
@@ -404,8 +454,8 @@ jobs:
       run:
         shell: msys2 {0}
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
-    - uses: msys2/setup-msys2@v2
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
+    - uses: msys2/setup-msys2@cc11e9188b693c2b100158c3322424c4cc1dadea # tag=v2.22.0
       with:
         msystem: ${{ matrix.msystem }}
         install: make diffutils
@@ -440,9 +490,9 @@ jobs:
         platform: [x64, Win32]
         configuration: [Release]
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: Add MSBuild to PATH
-      uses: microsoft/setup-msbuild@v1.3
+      uses: microsoft/setup-msbuild@6fb02220983dee41ce7ae257b6f4d8f9bf5ed4ce # tag=v2.0.0
     - name: Build and run tests
       working-directory: ${{env.GITHUB_WORKSPACE}}
       env:
@@ -461,8 +511,8 @@ jobs:
     runs-on: windows-latest
     steps:
     - run: git config --global core.autocrlf input
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
-    - uses: cygwin/cygwin-install-action@f5e0f048310c425e84bc789f493a828c6dc80a25 # tag=master
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
+    - uses: cygwin/cygwin-install-action@006ad0b0946ca6d0a3ea2d4437677fa767392401 # tag=master
       with:
         platform: x86_64
         packages: >-
@@ -483,31 +533,12 @@ jobs:
         make -C tests fuzzer &&
         ./tests/fuzzer.exe -v -T1m
 
-  intel-cet-compatibility:
-    runs-on: ubuntu-latest
-    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
-    - name: Build Zstd
-      run: |
-        make -j zstd V=1
-        readelf -n zstd
-    - name: Get Intel SDE
-      run: |
-        curl -LO https://downloadmirror.intel.com/684899/sde-external-9.0.0-2021-11-07-lin.tar.xz
-        tar xJvf sde-external-9.0.0-2021-11-07-lin.tar.xz
-    - name: Configure Permissions
-      run: |
-        echo 0 | sudo tee /proc/sys/kernel/yama/ptrace_scope
-    - name: Run Under SDE
-      run: |
-        sde-external-9.0.0-2021-11-07-lin/sde -cet -cet-raise 0 -cet-endbr-exe -cet-stderr -cet-abort -- ./zstd -b3
-
   pkg-config:
     runs-on: ubuntu-latest
     container:
       image: debian:testing
     steps:
-      - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+      - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
       - name: Install dependencies
         run: |
           apt -y update
@@ -522,7 +553,7 @@ jobs:
   versions-compatibility:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: Versions Compatibility Test
       run: |
         make -C tests versionsTest
@@ -530,7 +561,7 @@ jobs:
   clangbuild:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: make clangbuild
       run: |
         make clangbuild
@@ -538,7 +569,7 @@ jobs:
   clang-pgo:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: Build PGO Zstd with Clang
       env:
         CC: clang-14
@@ -550,7 +581,7 @@ jobs:
   gcc-pgo:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
     - name: Build PGO Zstd with GCC
       env:
         CC: gcc
@@ -558,10 +589,29 @@ jobs:
         make -C programs zstd-pgo
         ./programs/zstd -b
 
+  intel-cet-compatibility:
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
+    - name: Build Zstd
+      run: |
+        make -j zstd V=1
+        readelf -n zstd
+    - name: Get Intel SDE
+      run: |
+        curl -LO https://downloadmirror.intel.com/813591/sde-external-9.33.0-2024-01-07-lin.tar.xz
+        tar xJvf sde-external-9.33.0-2024-01-07-lin.tar.xz
+    - name: Configure Permissions
+      run: |
+        echo 0 | sudo tee /proc/sys/kernel/yama/ptrace_scope
+    - name: Run Under SDE
+      run: |
+        sde-external-9.33.0-2024-01-07-lin/sde -cet -cet-raise 0 -cet-endbr-exe -cet-stderr -cet-abort -- ./zstd -b3
+
+
+# Failing tests, for reference
 
-# For reference : icc tests
 # icc tests are currently failing on Github Actions, likely to issues during installation stage
-# To be fixed later
 #
 #  icc:
 #    name: icc-check
@@ -577,7 +627,7 @@ jobs:
 #        sudo add-apt-repository "deb https://apt.repos.intel.com/oneapi all main"
 #        sudo apt-get update
 #        sudo apt-get install -y intel-basekit intel-hpckit
-#    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+#    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v4.1.1
 #    - name: make check
 #      run: |
 #        make CC=/opt/intel/oneapi/compiler/latest/linux/bin/intel64/icc check

.github/workflows/nightly.yml

@@ -0,0 +1,65 @@
+name: facebook/zstd/nightly
+on:
+  schedule:
+  - cron: 0 0 * * *
+  push:
+    branches:
+    - release
+    - dev
+    - master
+permissions: read-all
+jobs:
+  regression-test:
+    runs-on: ubuntu-latest
+    services:
+      docker:
+        image: fbopensource/zstd-circleci-primary:0.0.1
+        options: --entrypoint /bin/bash
+    env:
+      CIRCLE_ARTIFACTS: "/tmp/circleci-artifacts"
+    steps:
+    - uses: actions/checkout@v4
+    - uses: actions/cache@v4
+      with:
+        key: regression-cache-{{ checksum "tests/regression/data.c" }}-v0
+        path: tests/regression/cache
+        restore-keys: regression-cache-{{ checksum "tests/regression/data.c" }}-v0
+    - uses: actions/upload-artifact@v4
+      with:
+        path: "/tmp/circleci-artifacts"
+    - name: Install Dependencies
+      run: |
+        sudo apt-get update
+        sudo apt-get install libcurl4-gnutls-dev
+    - name: Regression Test
+      run: |
+        make -C programs zstd
+        make -C tests/regression test
+        mkdir -p $CIRCLE_ARTIFACTS
+        ./tests/regression/test                     \
+            --cache  tests/regression/cache         \
+            --output $CIRCLE_ARTIFACTS/results.csv  \
+            --zstd   programs/zstd
+        echo "NOTE: The new results.csv is uploaded as an artifact to this job"
+        echo "      If this fails, go to the Artifacts pane in CircleCI, "
+        echo "      download /tmp/circleci-artifacts/results.csv, and if they "
+        echo "      are still good, copy it into the repo and commit it."
+        echo "> diff tests/regression/results.csv $CIRCLE_ARTIFACTS/results.csv"
+        diff tests/regression/results.csv $CIRCLE_ARTIFACTS/results.csv
+
+# Longer tests
+  #- make -C tests test-zstd-nolegacy && make clean
+  #- pyenv global 3.4.4; make -C tests versionsTest && make clean
+  #- make zlibwrapper         && make clean
+  #- gcc -v; make -C tests test32 MOREFLAGS="-I/usr/include/x86_64-linux-gnu" && make clean
+  #- make uasan               && make clean
+  #- make asan32              && make clean
+  #- make -C tests test32 CC=clang MOREFLAGS="-g -fsanitize=address -I/usr/include/x86_64-linux-gnu"
+# Valgrind tests
+  #- CFLAGS="-O1 -g" make -C zlibWrapper valgrindTest && make clean
+  #- make -C tests valgrindTest && make clean
+# ARM, AArch64, PowerPC, PowerPC64 tests
+  #- make ppctest             && make clean
+  #- make ppc64test           && make clean
+  #- make armtest             && make clean
+  #- make aarch64test         && make clean

.github/workflows/publish-release-artifacts.yml

@@ -5,21 +5,19 @@ on:
     types:
       - published
 
-permissions:
-  contents: read
+permissions: read-all
 
 jobs:
   publish-release-artifacts:
     permissions:
-      contents: read # to fetch code (actions/checkout)
-      actions: write # to attach binaries to release artifacts (skx/github-action-publish-binaries)
+      contents: write # to fetch code and upload artifacts
 
     runs-on: ubuntu-latest
     if: startsWith(github.ref, 'refs/tags/')
 
     steps:
       - name: Checkout
-        uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+        uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v3
 
       - name: Archive
         env:
@@ -68,7 +66,7 @@ jobs:
           fi
 
       - name: Publish
-        uses: skx/github-action-publish-binaries@release-2.0
+        uses: skx/github-action-publish-binaries@b9ca5643b2f1d7371a6cba7f35333f1461bbc703 # tag=release-2.0
         env:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
         with:

.github/workflows/scorecards.yml

@@ -27,12 +27,12 @@ jobs:
 
     steps:
       - name: "Checkout code"
-        uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+        uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v3
         with:
           persist-credentials: false
 
       - name: "Run analysis"
-        uses: ossf/scorecard-action@e38b1902ae4f44df626f11ba0734b14fb91f8f86 # tag=v2.1.2
+        uses: ossf/scorecard-action@0864cf19026789058feabb7e87baa5f140aac736 # tag=v2.3.1
         with:
           results_file: results.sarif
           results_format: sarif
@@ -51,7 +51,7 @@ jobs:
       # Upload the results as artifacts (optional). Commenting out will disable uploads of run results in SARIF
       # format to the repository Actions tab.
       - name: "Upload artifact"
-        uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce # tag=v3.1.2
+        uses: actions/upload-artifact@5d5d22a31266ced268874388b861e4b58bb5c2f3 # tag=v4.3.1
         with:
           name: SARIF file
           path: results.sarif
@@ -59,6 +59,6 @@ jobs:
 
       # Upload the results to GitHub's code scanning dashboard.
       - name: "Upload to code-scanning"
-        uses: github/codeql-action/upload-sarif@3ebbd71c74ef574dbc558c82f70e52732c8b44fe # tag=v2.2.1
+        uses: github/codeql-action/upload-sarif@3ab4101902695724f9365a384f86c1074d94e18c # tag=v3.24.7
         with:
           sarif_file: results.sarif

.github/workflows/windows-artifacts.yml

@@ -0,0 +1,58 @@
+name: windows-artifacts
+
+on:
+  push:
+    branches: [ test_artifacts, win_artifacts ]
+  release:
+    types:
+      - published
+
+permissions: read-all
+
+jobs:
+  windows-artifacts:
+    # see https://ariya.io/2020/07/on-github-actions-with-msys2
+    runs-on: windows-latest
+    # see https://github.com/msys2/setup-msys2
+    strategy:
+      matrix:
+        include:
+          - { msystem: mingw64, env: x86_64, ziparch: win64 }
+          - { msystem: mingw32, env: i686, ziparch: win32 }
+    defaults:
+      run:
+        shell: msys2 {0}
+    steps:
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # tag=v3
+    - uses: msys2/setup-msys2@5beef6d11f48bba68b9eb503e3adc60b23c0cc36 # tag=v2
+      with:
+        msystem: ${{ matrix.msystem }}
+        install: make zlib git p7zip mingw-w64-${{matrix.env}}-gcc
+        update: true
+
+    - name: display versions
+      run: |
+        make -v
+        cc -v
+
+    - name: Building zlib to static link
+      run: |
+        git clone --depth 1 --branch v1.2.11 https://github.com/madler/zlib
+        make -C zlib -f win32/Makefile.gcc libz.a
+
+    - name: Building zstd programs
+      run: |
+        CPPFLAGS=-I../zlib LDFLAGS=../zlib/libz.a make -j allzstd MOREFLAGS=-static V=1
+
+    - name: Create artifacts
+      run: |
+        ./lib/dll/example/build_package.bat
+        mv bin/ zstd-${{ github.ref_name }}-${{matrix.ziparch}}/
+        7z a -tzip -mx9 zstd-${{ github.ref_name }}-${{matrix.ziparch}}.zip zstd-${{ github.ref_name }}-${{matrix.ziparch}}/
+        cd ..
+
+    - name: Publish zstd-$VERSION-${{matrix.ziparch}}.zip
+      uses: actions/upload-artifact@5d5d22a31266ced268874388b861e4b58bb5c2f3 # tag=v4.3.1
+      with:
+        path: ${{ github.workspace }}/zstd-${{ github.ref_name }}-${{matrix.ziparch}}.zip
+        name: zstd-${{ github.ref_name }}-${{matrix.ziparch}}.zip

.gitignore

@@ -27,6 +27,8 @@ tmp*
 dictionary.
 dictionary
 NUL
+cmakebuild/
+install/
 
 # Build artefacts
 contrib/linux-kernel/linux/
@@ -37,11 +39,15 @@ buck-out/
 build-*
 *.gcda
 
+# IDE
+.clang_complete
+compile_flags.txt
+.clang-format
+
 # Other files
 .directory
 _codelite/
 _zstdbench/
-.clang_complete
 *.idea
 *.swp
 .DS_Store

.travis.yml

@@ -1,128 +0,0 @@
-# Travis CI is used to test platforms that github-actions currently doesn't support
-# without either self-hosting or some finnicky work-around. Also, some tests
-# are troublesome to migrate since GH Actions runs tests not in a tty.
-language: c
-
-git:
-  depth: 1
-
-branches:
-  only:
-  - dev
-  - release
-  - master
-  - travisTest
-
-addons:
-  apt:
-    update: true
-
-env:
-  global:
-    - FUZZERTEST=-T1mn
-      ZSTREAM_TESTTIME=-T1mn
-      DECODECORPUS_TESTTIME=-T1mn
-
-matrix:
-  fast_finish: true
-  include:
-    - name: S390X (big endian) + Fuzz test
-      dist: trusty
-      arch: s390x
-      script:
-        - FUZZER_FLAGS=--no-big-tests make -C tests fuzztest
-
-    - name: S390X (big endian) + Fuzz test + no intrinsics
-      dist: trusty
-      arch: s390x
-      script:
-        - MOREFLAGS="-DZSTD_NO_INTRINSICS" FUZZER_FLAGS=--no-big-tests make -C tests fuzztest
-
-    - name: arm64    # ~2.5 mn
-      os: linux
-      arch: arm64
-      script:
-        - make check
-
-    - name: arm64fuzz
-      os: linux
-      arch: arm64
-      script:
-        - make -C tests fuzztest
-
-    # TODO: migrate to GH Actions once newest clang staticanalyze warnings are fixed
-    - name: static analyzer scanbuild    # ~8mn
-      dist: trusty  # note : it's important to pin down a version of static analyzer, since different versions report different false positives
-      script:
-        - make staticAnalyze
-
-    # GH actions can't run this command on OS-X, non-tty issues
-    - name: OS-X make all lib
-      os: osx
-      script:
-        - make -C lib all
-
-    # Introduced to check compat with old toolchains, to prevent e.g. #1872
-    - name: ARM Build Test (on Trusty)
-      dist: trusty
-      script:
-        - make arminstall
-        - make armbuild
-
-    # check release number (release/new tag only)
-    - name: Tag-Specific Test
-      if: tag =~ ^v[0-9]\.[0-9]
-      script:
-        - make -C tests checkTag
-        - tests/checkTag "$TRAVIS_BRANCH"
-
-    - name: PPC64LE + Fuzz test  # ~13mn
-      arch: ppc64le
-      env:
-        - FUZZER_FLAGS=--no-big-tests
-        - MOREFLAGS="-static"
-      script:
-        - cat /proc/cpuinfo
-        - make -C tests fuzztest
-
-    # This test currently fails on GA specifically, for no obvious reason
-    # (it works fine on travisCI, and on local test platforms).
-    - name: Versions Compatibility Test   # ~6mn
-      script:
-        - make -C tests versionsTest
-
-    # meson dedicated test
-    - name: Focal (Meson + clang)    # ~15mn
-      dist: focal
-      language: cpp
-      compiler: clang
-      install:
-        - sudo apt-get install -qq liblz4-dev valgrind tree
-        - |
-          travis_retry curl -o ~/ninja.zip -L 'https://github.com/ninja-build/ninja/releases/download/v1.9.0/ninja-linux.zip' &&
-          unzip ~/ninja.zip -d ~/.local/bin
-        - |
-          travis_retry curl -o ~/get-pip.py -L 'https://bootstrap.pypa.io/pip/3.6/get-pip.py' &&
-          python3 ~/get-pip.py --user &&
-          pip3 install --user meson
-      script:
-        - |
-          meson setup \
-            --buildtype=debugoptimized \
-            -Db_lundef=false \
-            -Dauto_features=enabled \
-            -Dbin_programs=true \
-            -Dbin_tests=true \
-            -Dbin_contrib=true \
-            -Ddefault_library=both \
-             build/meson builddir
-        - pushd builddir
-        - ninja
-        - meson test --verbose --no-rebuild
-        - DESTDIR=./staging ninja install
-        - tree ./staging
-      after_failure:
-        - cat "$TRAVIS_BUILD_DIR"/builddir/meson-logs/testlog.txt
-
-  allow_failures:
-    - env: ALLOW_FAILURES=true

CHANGELOG

@@ -1,3 +1,59 @@
+V1.5.6 (Mar 2024)
+api: Promote `ZSTD_c_targetCBlockSize` to Stable API by @felixhandte
+api: new `ZSTD_d_maxBlockSize` experimental parameter, to reduce streaming decompression memory, by @terrelln
+perf: improve performance of param `ZSTD_c_targetCBlockSize`, by @Cyan4973
+perf: improved compression of arrays of integers at high compression, by @Cyan4973
+lib: reduce binary size with selective built-time exclusion, by @felixhandte
+lib: improved huffman speed on small data and linux kernel, by @terrelln
+lib: accept dictionaries with partial literal tables, by @terrelln
+lib: fix CCtx size estimation with external sequence producer, by @embg
+lib: fix corner case decoder behaviors, by @Cyan4973 and @aimuz
+lib: fix zdict prototype mismatch in static_only mode, by @ldv-alt
+lib: fix several bugs in magicless-format decoding, by @embg
+cli: add common compressed file types to `--exclude-compressed`` by @daniellerozenblit
+cli: fix mixing `-c` and `-o` commands with `--rm`, by @Cyan4973
+cli: fix erroneous exclusion of hidden files with `--output-dir-mirror` by @felixhandte
+cli: improved time accuracy on BSD, by @felixhandte
+cli: better errors on argument parsing, by @KapJI
+tests: better compatibility with older versions of `grep`, by @Cyan4973
+tests: lorem ipsum generator as default backup content, by @Cyan4973
+build: cmake improvements by @terrelln, @sighingnow, @gjasny, @JohanMabille, @Saverio976, @gruenich, @teo-tsirpanis
+build: bazel support, by @jondo2010
+build: fix cross-compiling for AArch64 with lld by @jcelerier
+build: fix Apple platform compatibility, by @nidhijaju
+build: fix Visual 2012 and lower compatibility, by @Cyan4973
+build: improve win32 support, by @DimitriPapadopoulos
+build: better C90 compliance for zlibWrapper, by @emaste
+port: make: fat binaries on macos, by @mredig
+port: ARM64EC compatibility for Windows, by @dunhor
+port: QNX support by @klausholstjacobsen
+port: MSYS2 and Cygwin makefile installation and test support, by @QBos07
+port: risc-v support validation in CI, by @Cyan4973
+port: sparc64 support validation in CI, by @Cyan4973
+port: AIX compatibility, by @likema
+port: HP-UX compatibility, by @likema
+doc: Improved specification accuracy, by @elasota
+bug: Fix and deprecate ZSTD_generateSequences (#3981)
+
+v1.5.5 (Apr 2023)
+fix: fix rare corruption bug affecting the high compression mode, reported by @danlark1 (#3517, @terrelln)
+perf: improve mid-level compression speed (#3529, #3533, #3543, @yoniko and #3552, @terrelln)
+lib: deprecated bufferless block-level API (#3534) by @terrelln
+cli: mmap large dictionaries to save memory, by @daniellerozenblit
+cli: improve speed of --patch-from mode (~+50%) (#3545) by @daniellerozenblit
+cli: improve i/o speed (~+10%) when processing lots of small files (#3479) by @felixhandte
+cli: zstd no longer crashes when requested to write into write-protected directory (#3541) by @felixhandte
+cli: fix decompression into block device using -o, reported by @georgmu (#3583)
+build: fix zstd CLI compiled with lzma support but not zlib support (#3494) by @Hello71
+build: fix cmake does no longer require 3.18 as minimum version (#3510) by @kou
+build: fix MSVC+ClangCL linking issue (#3569) by @tru
+build: fix zstd-dll, version of zstd CLI that links to the dynamic library (#3496) by @yoniko
+build: fix MSVC warnings (#3495) by @embg
+doc: updated zstd specification to clarify corner cases, by @Cyan4973
+doc: document how to create fat binaries for macos (#3568) by @rickmark
+misc: improve seekable format ingestion speed (~+100%) for very small chunk sizes (#3544) by @Cyan4973
+misc: tests/fullbench can benchmark multiple files (#3516) by @dloidolt
+
 v1.5.4 (Feb 2023)
 perf: +20% faster huffman decompression for targets that can't compile x64 assembly (#3449, @terrelln)
 perf: up to +10% faster streaming compression at levels 1-2 (#3114, @embg)
@@ -79,7 +135,7 @@ build: support for m68k (Motorola 68000's), by @cyan4973
 build: improved AIX support, by @Helflym
 build: improved meson unofficial build, by @eli-schwartz
 cli : custom memory limit when training dictionary (#2925), by @embg
-cli : report advanced parameters information when compressing in very verbose mode (``-vv`), by @Svetlitski-FB
+cli : report advanced parameters information when compressing in very verbose mode (`-vv`), by @Svetlitski-FB
 
 v1.5.0  (May 11, 2021)
 api: Various functions promoted from experimental to stable API: (#2579-2581, @senhuang42)
@@ -146,7 +202,7 @@ api: Add Function to Generate Skippable Frame (#2439, @senhuang42)
 perf: New Algorithms for the Long Distance Matcher (#2483, @mpu)
 perf: Performance Improvements for Long Distance Matcher (#2464, @mpu)
 perf: Don't Shrink Window Log when Streaming with a Dictionary (#2451, @terrelln)
-cli: Fix `--output-dir-mirror`'s Rejection of `..`-Containing Paths (#2512, @felixhandte)
+cli: Fix `--output-dir-mirror` rejection of `..` -containing paths (#2512, @felixhandte)
 cli: Allow Input From Console When `-f`/`--force` is Passed (#2466, @felixhandte)
 cli: Improve Help Message (#2500, @senhuang42)
 tests: Remove Flaky Tests (#2455, #2486, #2445, @Cyan4973)

CONTRIBUTING.md

@@ -171,8 +171,8 @@ who want earlier signal.
 | Cirrus CI | Used for testing on FreeBSD                                                                                | https://github.com/marketplace/cirrus-ci/                                                                                                              | `.cirrus.yml`          |
 | Circle CI | Historically was used to provide faster signal,<br/> but we may be able to migrate these to Github Actions | https://circleci.com/docs/2.0/getting-started/#setting-up-circleci <br> https://youtu.be/Js3hMUsSZ2c <br> https://circleci.com/docs/2.0/enable-checks/ | `.circleci/config.yml` |
 
-Note: the instructions linked above mostly cover how to set up a repository with CI from scratch. 
-The general idea should be the same for setting up CI on your fork of zstd, but you may have to 
+Note: the instructions linked above mostly cover how to set up a repository with CI from scratch.
+The general idea should be the same for setting up CI on your fork of zstd, but you may have to
 follow slightly different steps. In particular, please ignore any instructions related to setting up
 config files (since zstd already has configs for each of these services).
 
@@ -216,7 +216,7 @@ will typically not be stable enough to obtain reliable benchmark results. If you
 hands on a desktop, this is usually a better scenario.
 
 Of course, benchmarking can be done on non-hyper-stable machines as well. You will just have to
-do a little more work to ensure that you are in fact measuring the changes you've made not and
+do a little more work to ensure that you are in fact measuring the changes you've made and not
 noise. Here are some things you can do to make your benchmarks more stable:
 
 1. The most simple thing you can do to drastically improve the stability of your benchmark is

Makefile

@@ -145,13 +145,13 @@ clean:
 	$(Q)$(MAKE) -C contrib/largeNbDicts $@ > $(VOID)
 	$(Q)$(MAKE) -C contrib/externalSequenceProducer $@ > $(VOID)
 	$(Q)$(RM) zstd$(EXT) zstdmt$(EXT) tmp*
-	$(Q)$(RM) -r lz4
+	$(Q)$(RM) -r lz4 cmakebuild install
 	@echo Cleaning completed
 
 #------------------------------------------------------------------------------
 # make install is validated only for Linux, macOS, Hurd and some BSD targets
 #------------------------------------------------------------------------------
-ifneq (,$(filter $(shell uname),Linux Darwin GNU/kFreeBSD GNU OpenBSD FreeBSD DragonFly NetBSD MSYS_NT Haiku AIX))
+ifneq (,$(filter $(shell uname),Linux Darwin GNU/kFreeBSD GNU OpenBSD FreeBSD DragonFly NetBSD MSYS_NT CYGWIN_NT Haiku AIX))
 
 HOST_OS = POSIX
 
@@ -197,18 +197,27 @@ uninstall:
 travis-install:
 	$(MAKE) install PREFIX=~/install_test_dir
 
+.PHONY: clangbuild-darwin-fat
+clangbuild-darwin-fat: clean
+	clang -v
+	CXX=clang++ CC=clang CFLAGS="-Werror -Wconversion -Wno-sign-conversion -Wdocumentation -arch arm64" $(MAKE) zstd-release
+	mv programs/zstd programs/zstd_arm64
+	CXX=clang++ CC=clang CFLAGS="-Werror -Wconversion -Wno-sign-conversion -Wdocumentation -arch x86_64" $(MAKE) zstd-release
+	mv programs/zstd programs/zstd_x64
+	lipo -create programs/zstd_x64 programs/zstd_arm64 -output programs/zstd
+
 .PHONY: gcc5build gcc6build gcc7build clangbuild m32build armbuild aarch64build ppcbuild ppc64build
 gcc5build: clean
 	gcc-5 -v
-	CC=gcc-5 $(MAKE) all MOREFLAGS="-Werror"
+	CC=gcc-5 $(MAKE) all MOREFLAGS="-Werror $(MOREFLAGS)"
 
 gcc6build: clean
 	gcc-6 -v
-	CC=gcc-6 $(MAKE) all MOREFLAGS="-Werror"
+	CC=gcc-6 $(MAKE) all MOREFLAGS="-Werror $(MOREFLAGS)"
 
 gcc7build: clean
 	gcc-7 -v
-	CC=gcc-7 $(MAKE) all MOREFLAGS="-Werror"
+	CC=gcc-7 $(MAKE) all MOREFLAGS="-Werror $(MOREFLAGS)"
 
 clangbuild: clean
 	clang -v
@@ -232,17 +241,17 @@ ppc64build: clean
 
 .PHONY: armfuzz aarch64fuzz ppcfuzz ppc64fuzz
 armfuzz: clean
-	CC=arm-linux-gnueabi-gcc QEMU_SYS=qemu-arm-static MOREFLAGS="-static" FUZZER_FLAGS=--no-big-tests $(MAKE) -C $(TESTDIR) fuzztest
+	CC=arm-linux-gnueabi-gcc QEMU_SYS=qemu-arm-static MOREFLAGS="-static $(MOREFLAGS)" FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)" $(MAKE) -C $(TESTDIR) fuzztest
 
 aarch64fuzz: clean
 	ld -v
-	CC=aarch64-linux-gnu-gcc QEMU_SYS=qemu-aarch64-static MOREFLAGS="-static" FUZZER_FLAGS=--no-big-tests $(MAKE) -C $(TESTDIR) fuzztest
+	CC=aarch64-linux-gnu-gcc QEMU_SYS=qemu-aarch64-static MOREFLAGS="-static $(MOREFLAGS)" FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)" $(MAKE) -C $(TESTDIR) fuzztest
 
 ppcfuzz: clean
-	CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc-static MOREFLAGS="-static" FUZZER_FLAGS=--no-big-tests $(MAKE) -C $(TESTDIR) fuzztest
+	CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc-static MOREFLAGS="-static $(MOREFLAGS)" FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)" $(MAKE) -C $(TESTDIR) fuzztest
 
 ppc64fuzz: clean
-	CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc64-static MOREFLAGS="-m64 -static" FUZZER_FLAGS=--no-big-tests $(MAKE) -C $(TESTDIR) fuzztest
+	CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc64-static MOREFLAGS="-m64 -static $(MOREFLAGS)" FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)" $(MAKE) -C $(TESTDIR) fuzztest
 
 .PHONY: cxxtest gcc5test gcc6test armtest aarch64test ppctest ppc64test
 cxxtest: CXXFLAGS += -Wall -Wextra -Wundef -Wshadow -Wcast-align -Werror
@@ -251,34 +260,34 @@ cxxtest: clean
 
 gcc5test: clean
 	gcc-5 -v
-	$(MAKE) all CC=gcc-5 MOREFLAGS="-Werror"
+	$(MAKE) all CC=gcc-5 MOREFLAGS="-Werror $(MOREFLAGS)"
 
 gcc6test: clean
 	gcc-6 -v
-	$(MAKE) all CC=gcc-6 MOREFLAGS="-Werror"
+	$(MAKE) all CC=gcc-6 MOREFLAGS="-Werror $(MOREFLAGS)"
 
 armtest: clean
 	$(MAKE) -C $(TESTDIR) datagen   # use native, faster
-	$(MAKE) -C $(TESTDIR) test CC=arm-linux-gnueabi-gcc QEMU_SYS=qemu-arm-static ZSTDRTTEST= MOREFLAGS="-Werror -static" FUZZER_FLAGS=--no-big-tests
+	$(MAKE) -C $(TESTDIR) test CC=arm-linux-gnueabi-gcc QEMU_SYS=qemu-arm-static ZSTDRTTEST= MOREFLAGS="-Werror -static $(MOREFLAGS)" FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)"
 
 aarch64test:
 	$(MAKE) -C $(TESTDIR) datagen   # use native, faster
-	$(MAKE) -C $(TESTDIR) test CC=aarch64-linux-gnu-gcc QEMU_SYS=qemu-aarch64-static ZSTDRTTEST= MOREFLAGS="-Werror -static" FUZZER_FLAGS=--no-big-tests
+	$(MAKE) -C $(TESTDIR) test CC=aarch64-linux-gnu-gcc QEMU_SYS=qemu-aarch64-static ZSTDRTTEST= MOREFLAGS="-Werror -static $(MOREFLAGS)" FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)"
 
 ppctest: clean
 	$(MAKE) -C $(TESTDIR) datagen   # use native, faster
-	$(MAKE) -C $(TESTDIR) test CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc-static ZSTDRTTEST= MOREFLAGS="-Werror -Wno-attributes -static" FUZZER_FLAGS=--no-big-tests
+	$(MAKE) -C $(TESTDIR) test CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc-static ZSTDRTTEST= MOREFLAGS="-Werror -Wno-attributes -static $(MOREFLAGS)" FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)"
 
 ppc64test: clean
 	$(MAKE) -C $(TESTDIR) datagen   # use native, faster
-	$(MAKE) -C $(TESTDIR) test CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc64-static ZSTDRTTEST= MOREFLAGS="-m64 -static" FUZZER_FLAGS=--no-big-tests
+	$(MAKE) -C $(TESTDIR) test CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc64-static ZSTDRTTEST= MOREFLAGS="-m64 -static $(MOREFLAGS)" FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)"
 
 .PHONY: arm-ppc-compilation
 arm-ppc-compilation:
-	$(MAKE) -C $(PRGDIR) clean zstd CC=arm-linux-gnueabi-gcc QEMU_SYS=qemu-arm-static ZSTDRTTEST= MOREFLAGS="-Werror -static"
-	$(MAKE) -C $(PRGDIR) clean zstd CC=aarch64-linux-gnu-gcc QEMU_SYS=qemu-aarch64-static ZSTDRTTEST= MOREFLAGS="-Werror -static"
-	$(MAKE) -C $(PRGDIR) clean zstd CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc-static ZSTDRTTEST= MOREFLAGS="-Werror -Wno-attributes -static"
-	$(MAKE) -C $(PRGDIR) clean zstd CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc64-static ZSTDRTTEST= MOREFLAGS="-m64 -static"
+	$(MAKE) -C $(PRGDIR) clean zstd CC=arm-linux-gnueabi-gcc QEMU_SYS=qemu-arm-static ZSTDRTTEST= MOREFLAGS="-Werror -static $(MOREFLAGS)"
+	$(MAKE) -C $(PRGDIR) clean zstd CC=aarch64-linux-gnu-gcc QEMU_SYS=qemu-aarch64-static ZSTDRTTEST= MOREFLAGS="-Werror -static $(MOREFLAGS)"
+	$(MAKE) -C $(PRGDIR) clean zstd CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc-static ZSTDRTTEST= MOREFLAGS="-Werror -Wno-attributes -static $(MOREFLAGS)"
+	$(MAKE) -C $(PRGDIR) clean zstd CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc64-static ZSTDRTTEST= MOREFLAGS="-m64 -static $(MOREFLAGS)"
 
 regressiontest:
 	$(MAKE) -C $(FUZZDIR) regressiontest
@@ -308,31 +317,32 @@ update_regressionResults:
 # run UBsan with -fsanitize-recover=pointer-overflow
 # this only works with recent compilers such as gcc 8+
 usan: clean
-	$(MAKE) test CC=clang MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize-recover=pointer-overflow -fsanitize=undefined -Werror"
+	$(MAKE) test CC=clang MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize=undefined -Werror $(MOREFLAGS)"
 
 asan: clean
-	$(MAKE) test CC=clang MOREFLAGS="-g -fsanitize=address -Werror"
+	$(MAKE) test CC=clang MOREFLAGS="-g -fsanitize=address -Werror $(MOREFLAGS)"
 
 asan-%: clean
-	LDFLAGS=-fuse-ld=gold MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize=address -Werror" $(MAKE) -C $(TESTDIR) $*
+	LDFLAGS=-fuse-ld=gold MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize=address -Werror $(MOREFLAGS)" $(MAKE) -C $(TESTDIR) $*
 
 msan: clean
-	$(MAKE) test CC=clang MOREFLAGS="-g -fsanitize=memory -fno-omit-frame-pointer -Werror" HAVE_LZMA=0   # datagen.c fails this test for no obvious reason
+	$(MAKE) test CC=clang MOREFLAGS="-g -fsanitize=memory -fno-omit-frame-pointer -Werror $(MOREFLAGS)" HAVE_LZMA=0   # datagen.c fails this test for no obvious reason
 
-msan-%: clean
-	LDFLAGS=-fuse-ld=gold MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize=memory -fno-omit-frame-pointer -Werror" FUZZER_FLAGS=--no-big-tests $(MAKE) -C $(TESTDIR) HAVE_LZMA=0 $*
+msan-%:
+	$(MAKE) clean
+	LDFLAGS=-fuse-ld=gold MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize=memory -fno-omit-frame-pointer -Werror $(MOREFLAGS)" FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)" $(MAKE) -j -C $(TESTDIR) HAVE_LZMA=0 $*
 
 asan32: clean
-	$(MAKE) -C $(TESTDIR) test32 CC=clang MOREFLAGS="-g -fsanitize=address"
+	$(MAKE) -C $(TESTDIR) test32 CC=clang MOREFLAGS="-g -fsanitize=address $(MOREFLAGS)"
 
 uasan: clean
-	$(MAKE) test CC=clang MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize-recover=pointer-overflow -fsanitize=address,undefined -Werror"
+	$(MAKE) test CC=clang MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize=address,undefined -Werror $(MOREFLAGS)"
 
 uasan-%: clean
-	LDFLAGS=-fuse-ld=gold MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize-recover=pointer-overflow -fsanitize=address,undefined -Werror" $(MAKE) -C $(TESTDIR) $*
+	LDFLAGS=-fuse-ld=gold MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize=address,undefined -Werror $(MOREFLAGS)" $(MAKE) -C $(TESTDIR) $*
 
 tsan-%: clean
-	LDFLAGS=-fuse-ld=gold MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize=thread -Werror" $(MAKE) -C $(TESTDIR) $* FUZZER_FLAGS=--no-big-tests
+	LDFLAGS=-fuse-ld=gold MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize=thread -Werror $(MOREFLAGS)" $(MAKE) -C $(TESTDIR) $* FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)"
 
 .PHONY: apt-install
 apt-install:
@@ -380,28 +390,32 @@ lz4install:
 endif
 
 
-CMAKE_PARAMS = -DZSTD_BUILD_CONTRIB:BOOL=ON -DZSTD_BUILD_STATIC:BOOL=ON -DZSTD_BUILD_TESTS:BOOL=ON -DZSTD_ZLIB_SUPPORT:BOOL=ON -DZSTD_LZMA_SUPPORT:BOOL=ON -DCMAKE_BUILD_TYPE=Release
-
 ifneq (,$(filter MSYS%,$(shell uname)))
 HOST_OS = MSYS
-CMAKE_PARAMS = -G"MSYS Makefiles" -DCMAKE_BUILD_TYPE=Debug -DZSTD_MULTITHREAD_SUPPORT:BOOL=OFF -DZSTD_BUILD_STATIC:BOOL=ON -DZSTD_BUILD_TESTS:BOOL=ON
 endif
 
 #------------------------------------------------------------------------
 # target specific tests
 #------------------------------------------------------------------------
 ifneq (,$(filter $(HOST_OS),MSYS POSIX))
-.PHONY: cmakebuild c89build gnu90build c99build gnu99build c11build bmix64build bmix32build bmi32build staticAnalyze
-cmakebuild:
-	cmake --version
-	$(RM) -r $(BUILDIR)/cmake/build
-	$(MKDIR) $(BUILDIR)/cmake/build
-	cd $(BUILDIR)/cmake/build; cmake -DCMAKE_INSTALL_PREFIX:PATH=~/install_test_dir $(CMAKE_PARAMS) ..
-	$(MAKE) -C $(BUILDIR)/cmake/build -j4;
-	$(MAKE) -C $(BUILDIR)/cmake/build install;
-	$(MAKE) -C $(BUILDIR)/cmake/build uninstall;
-	cd $(BUILDIR)/cmake/build; ctest -V -L Medium
 
+CMAKE ?= cmake
+CMAKE_PARAMS = -DZSTD_BUILD_CONTRIB:BOOL=ON -DZSTD_BUILD_STATIC:BOOL=ON -DZSTD_BUILD_TESTS:BOOL=ON -DZSTD_ZLIB_SUPPORT:BOOL=ON -DZSTD_LZMA_SUPPORT:BOOL=ON
+
+ifneq (,$(filter MSYS%,$(shell uname)))
+CMAKE_PARAMS = -G"MSYS Makefiles" -DZSTD_MULTITHREAD_SUPPORT:BOOL=OFF -DZSTD_BUILD_STATIC:BOOL=ON -DZSTD_BUILD_TESTS:BOOL=ON
+endif
+
+.PHONY: cmakebuild
+cmakebuild:
+	$(CMAKE) --version
+	$(RM) -r cmakebuild install
+	$(MKDIR) cmakebuild install
+	cd cmakebuild; $(CMAKE) -Wdev -DCMAKE_BUILD_TYPE=Debug -DCMAKE_C_FLAGS="-Werror -O0" -DCMAKE_INSTALL_PREFIX=install $(CMAKE_PARAMS) ../build/cmake
+	$(CMAKE) --build cmakebuild --target install -- -j V=1
+	cd cmakebuild; ctest -V -L Medium
+
+.PHONY: c89build gnu90build c99build gnu99build c11build bmix64build bmix32build bmi32build staticAnalyze
 c89build: clean
 	$(CC) -v
 	CFLAGS="-std=c89 -Werror -Wno-attributes -Wpedantic -Wno-long-long -Wno-variadic-macros -O0" $(MAKE) lib zstd

README.md

@@ -5,7 +5,7 @@ targeting real-time compression scenarios at zlib-level and better compression r
 It's backed by a very fast entropy stage, provided by [Huff0 and FSE library](https://github.com/Cyan4973/FiniteStateEntropy).
 
 Zstandard's format is stable and documented in [RFC8878](https://datatracker.ietf.org/doc/html/rfc8878). Multiple independent implementations are already available.
-This repository represents the reference implementation, provided as an open-source dual [BSD](LICENSE) and [GPLv2](COPYING) licensed **C** library,
+This repository represents the reference implementation, provided as an open-source dual [BSD](LICENSE) OR [GPLv2](COPYING) licensed **C** library,
 and a command line utility producing and decoding `.zst`, `.gz`, `.xz` and `.lz4` files.
 Should your project require another programming language,
 a list of known ports and bindings is provided on [Zstandard homepage](https://facebook.github.io/zstd/#other-languages).
@@ -13,15 +13,12 @@ a list of known ports and bindings is provided on [Zstandard homepage](https://f
 **Development branch status:**
 
 [![Build Status][travisDevBadge]][travisLink]
-[![Build status][AppveyorDevBadge]][AppveyorLink]
 [![Build status][CircleDevBadge]][CircleLink]
 [![Build status][CirrusDevBadge]][CirrusLink]
 [![Fuzzing Status][OSSFuzzBadge]][OSSFuzzLink]
 
 [travisDevBadge]: https://api.travis-ci.com/facebook/zstd.svg?branch=dev "Continuous Integration test suite"
 [travisLink]: https://travis-ci.com/facebook/zstd
-[AppveyorDevBadge]: https://ci.appveyor.com/api/projects/status/xt38wbdxjk5mrbem/branch/dev?svg=true "Windows test suite"
-[AppveyorLink]: https://ci.appveyor.com/project/YannCollet/zstd-p0yf0
 [CircleDevBadge]: https://circleci.com/gh/facebook/zstd/tree/dev.svg?style=shield "Short test suite"
 [CircleLink]: https://circleci.com/gh/facebook/zstd
 [CirrusDevBadge]: https://api.cirrus-ci.com/github/facebook/zstd.svg?branch=dev
@@ -154,6 +151,18 @@ to create `zstd` binary, and `libzstd` dynamic and static libraries.
 
 By default, `CMAKE_BUILD_TYPE` is set to `Release`.
 
+#### Support for Fat (Universal2) Output
+
+`zstd` can be built and installed with support for both Apple Silicon (M1/M2) as well as Intel by using CMake's Universal2 support.
+To perform a Fat/Universal2 build and install use the following commands:
+
+```bash
+cmake -B build-cmake-debug -S build/cmake -G Ninja -DCMAKE_OSX_ARCHITECTURES="x86_64;x86_64h;arm64"
+cd build-cmake-debug
+ninja
+sudo ninja install
+```
+
 ### Meson
 
 A Meson project is provided within [`build/meson`](build/meson). Follow
@@ -189,6 +198,10 @@ Going into `build` directory, you will find additional possibilities:
 You can build the zstd binary via buck by executing: `buck build programs:zstd` from the root of the repo.
 The output binary will be in `buck-out/gen/programs/`.
 
+### Bazel
+
+You easily can integrate zstd into your Bazel project by using the module hosted on the [Bazel Central Repository](https://registry.bazel.build/modules/zstd).
+
 ## Testing
 
 You can run quick local smoke tests by running `make check`.
@@ -204,7 +217,7 @@ Zstandard is considered safe for production environments.
 
 ## License
 
-Zstandard is dual-licensed under [BSD](LICENSE) and [GPLv2](COPYING).
+Zstandard is dual-licensed under [BSD](LICENSE) OR [GPLv2](COPYING).
 
 ## Contributing
 

SECURITY.md

@@ -0,0 +1,15 @@
+# Reporting and Fixing Security Issues
+
+Please do not open GitHub issues or pull requests - this makes the problem immediately visible to everyone, including malicious actors. Security issues in this open source project can be safely reported via the Meta Bug Bounty program:
+
+https://www.facebook.com/whitehat
+
+Meta's security team will triage your report and determine whether or not is it eligible for a bounty under our program.
+
+# Receiving Vulnerability Notifications
+
+In the case that a significant security vulnerability is reported to us or discovered by us---without being publicly known---we will, at our discretion, notify high-profile, high-exposure users of Zstandard ahead of our public disclosure of the issue and associated fix.
+
+If you believe your project would benefit from inclusion in this list, please reach out to one of the maintainers.
+
+<!-- Note to maintainers: this list is kept [here](https://fburl.com/wiki/cgc1l62x). -->

appveyor.yml

@@ -1,205 +0,0 @@
-# Following tests are run _only_ on `release` branch
-# and on selected feature branch named `appveyorTest` or `visual*`
-
--
-  version: 1.0.{build}
-  branches:
-    only:
-    - release
-    - master
-    - /appveyor*/
-    - /visual*/
-  environment:
-    matrix:
-    - COMPILER: "gcc"
-      HOST:     "mingw"
-      PLATFORM: "x64"
-      SCRIPT:   "make allzstd MOREFLAGS=-static"
-      ARTIFACT: "true"
-      BUILD:    "true"
-    - COMPILER: "gcc"
-      HOST:     "mingw"
-      PLATFORM: "x86"
-      SCRIPT:   "make allzstd MOREFLAGS=-static"
-      ARTIFACT: "true"
-      BUILD:    "true"
-
-    - COMPILER: "clang-cl"
-      HOST:     "cmake-visual"
-      PLATFORM: "x64"
-      CONFIGURATION: "Release"
-      CMAKE_GENERATOR: "Visual Studio 15 2017"
-      CMAKE_GENERATOR_PLATFORM: "x64"
-      CMAKE_GENERATOR_TOOLSET: "LLVM"
-      APPVEYOR_BUILD_WORKER_IMAGE: "Visual Studio 2017"
-
-  install:
-  - ECHO Installing %COMPILER% %PLATFORM% %CONFIGURATION%
-  - SET PATH_ORIGINAL=%PATH%
-  - if [%HOST%]==[mingw] (
-      SET "PATH_MINGW32=C:\mingw-w64\i686-6.3.0-posix-dwarf-rt_v5-rev1\mingw32\bin" &&
-      SET "PATH_MINGW64=C:\mingw-w64\x86_64-6.3.0-posix-seh-rt_v5-rev1\mingw64\bin" &&
-      COPY C:\msys64\usr\bin\make.exe C:\mingw-w64\i686-6.3.0-posix-dwarf-rt_v5-rev1\mingw32\bin\make.exe &&
-      COPY C:\msys64\usr\bin\make.exe C:\mingw-w64\x86_64-6.3.0-posix-seh-rt_v5-rev1\mingw64\bin\make.exe
-    )
-  - IF [%HOST%]==[visual] IF [%PLATFORM%]==[x64] (
-      SET ADDITIONALPARAM=/p:LibraryPath="C:\Program Files\Microsoft SDKs\Windows\v7.1\lib\x64;c:\Program Files (x86)\Microsoft Visual Studio 10.0\VC\lib\amd64;C:\Program Files (x86)\Microsoft Visual Studio 10.0\;C:\Program Files (x86)\Microsoft Visual Studio 10.0\lib\amd64;"
-    )
-
-  build_script:
-  - if [%HOST%]==[mingw] (
-      ( if [%PLATFORM%]==[x64] (
-        SET "PATH=%PATH_MINGW64%;%PATH_ORIGINAL%"
-      ) else if [%PLATFORM%]==[x86] (
-        SET "PATH=%PATH_MINGW32%;%PATH_ORIGINAL%"
-      ) )
-    )
-  - if [%HOST%]==[mingw] if [%BUILD%]==[true] (
-      make -v &&
-      sh -c "%COMPILER% -v" &&
-      ECHO Building zlib to static link &&
-      SET "CC=%COMPILER%" &&
-      sh -c "cd .. && git clone --depth 1 --branch v1.2.11 https://github.com/madler/zlib" &&
-      sh -c "cd ../zlib && make -f win32/Makefile.gcc libz.a"
-      ECHO Building zstd &&
-      SET "CPPFLAGS=-I../../zlib" &&
-      SET "LDFLAGS=../../zlib/libz.a" &&
-      sh -c "%SCRIPT%" &&
-      ( if [%COMPILER%]==[gcc] if [%ARTIFACT%]==[true]
-          ECHO Creating artifacts &&
-          ECHO %cd% &&
-          lib\dll\example\build_package.bat &&
-          make -C programs DEBUGFLAGS= clean zstd &&
-          cd programs\ && 7z a -tzip -mx9 zstd-win-binary-%PLATFORM%.zip zstd.exe &&
-          appveyor PushArtifact zstd-win-binary-%PLATFORM%.zip &&
-          cp zstd.exe ..\bin\zstd.exe &&
-          git clone --depth 1 --branch release https://github.com/facebook/zstd &&
-          cd zstd &&
-          git archive --format=tar release -o zstd-src.tar &&
-          ..\zstd -19 zstd-src.tar &&
-          appveyor PushArtifact zstd-src.tar.zst &&
-          certUtil -hashfile zstd-src.tar.zst SHA256 > zstd-src.tar.zst.sha256.sig &&
-          appveyor PushArtifact zstd-src.tar.zst.sha256.sig &&
-          cd ..\..\bin\ &&
-          7z a -tzip -mx9 zstd-win-release-%PLATFORM%.zip * &&
-          appveyor PushArtifact zstd-win-release-%PLATFORM%.zip
-      )
-    )
-  - if [%HOST%]==[cmake-visual] (
-      ECHO *** &&
-      ECHO *** Building %CMAKE_GENERATOR% ^(%CMAKE_GENERATOR_TOOLSET%^) %PLATFORM%\%CONFIGURATION% &&
-      PUSHD build\cmake &&
-      cmake -DBUILD_TESTING=ON . &&
-      cmake --build . --config %CONFIGURATION% -j4 &&
-      POPD &&
-      ECHO ***
-    )
-
-  test_script:
-  - ECHO Testing %COMPILER% %PLATFORM% %CONFIGURATION%
-  - SET "CC=gcc"
-  - SET "CXX=g++"
-  - if [%TEST%]==[cmake] (
-      mkdir build\cmake\build &&
-      cd build\cmake\build &&
-      SET FUZZERTEST=-T2mn &&
-      SET ZSTREAM_TESTTIME=-T2mn &&
-      cmake -G "Visual Studio 14 2015 Win64" .. &&
-      cd ..\..\.. &&
-      make clean
-    )
-
-
-# The following tests are for regular pushes
-# into `dev` or some feature branch
-# There run less tests, for shorter feedback loop
-
--
-  version: 1.0.{build}
-  environment:
-    matrix:
-    - COMPILER: "visual"
-      HOST:     "visual"
-      PLATFORM: "x64"
-      CONFIGURATION: "Debug"
-    - COMPILER: "visual"
-      HOST:     "visual"
-      PLATFORM: "Win32"
-      CONFIGURATION: "Debug"
-    - COMPILER: "visual"
-      HOST:     "visual"
-      PLATFORM: "x64"
-      CONFIGURATION: "Release"
-    - COMPILER: "visual"
-      HOST:     "visual"
-      PLATFORM: "Win32"
-      CONFIGURATION: "Release"
-
-    - COMPILER: "gcc"
-      HOST:     "cygwin"
-      PLATFORM: "x64"
-
-    - COMPILER: "clang-cl"
-      HOST:     "cmake-visual"
-      PLATFORM: "x64"
-      CONFIGURATION: "Release"
-      CMAKE_GENERATOR: "Visual Studio 15 2017"
-      CMAKE_GENERATOR_PLATFORM: "x64"
-      CMAKE_GENERATOR_TOOLSET: "LLVM"
-      APPVEYOR_BUILD_WORKER_IMAGE: "Visual Studio 2017"
-
-  install:
-  - ECHO Installing %COMPILER% %PLATFORM% %CONFIGURATION%
-  - SET PATH_ORIGINAL=%PATH%
-  - if [%HOST%]==[cygwin] (
-      ECHO Installing Cygwin Packages &&
-      C:\cygwin64\setup-x86_64.exe -qnNdO -R "C:\cygwin64" -g -P ^
-        gcc,^
-        cmake,^
-        make
-    )
-  - IF [%HOST%]==[visual] IF [%PLATFORM%]==[x64] (
-      SET ADDITIONALPARAM=/p:LibraryPath="C:\Program Files\Microsoft SDKs\Windows\v7.1\lib\x64;c:\Program Files (x86)\Microsoft Visual Studio 10.0\VC\lib\amd64;C:\Program Files (x86)\Microsoft Visual Studio 10.0\;C:\Program Files (x86)\Microsoft Visual Studio 10.0\lib\amd64;"
-    )
-
-  build_script:
-  - ECHO Building %COMPILER% %PLATFORM% %CONFIGURATION%
-  - if [%HOST%]==[cygwin] (
-      set CHERE_INVOKING=yes &&
-      set CC=%COMPILER% &&
-      C:\cygwin64\bin\bash --login -c "
-        set -e;
-        cd build/cmake;
-        CFLAGS='-Werror' cmake -G 'Unix Makefiles' -DCMAKE_BUILD_TYPE=Debug -DZSTD_BUILD_TESTS:BOOL=ON -DZSTD_FUZZER_FLAGS=-T20s -DZSTD_ZSTREAM_FLAGS=-T20s -DZSTD_FULLBENCH_FLAGS=-i0 .;
-        make VERBOSE=1 -j;
-        ctest -V -L Medium;
-      "
-    )
-  - if [%HOST%]==[cmake-visual] (
-      ECHO *** &&
-      ECHO *** Building %CMAKE_GENERATOR% ^(%CMAKE_GENERATOR_TOOLSET%^) %PLATFORM%\%CONFIGURATION% &&
-      PUSHD build\cmake &&
-      cmake -DBUILD_TESTING=ON . &&
-      cmake --build . --config %CONFIGURATION% -j4 &&
-      POPD &&
-      ECHO ***
-    )
-  - if [%HOST%]==[visual] (
-      ECHO *** &&
-      ECHO *** Building Visual Studio 2012 %PLATFORM%\%CONFIGURATION% &&
-      ECHO *** &&
-      msbuild "build\VS2010\zstd.sln" /m /verbosity:minimal /property:PlatformToolset=v110 /p:ForceImportBeforeCppTargets=%APPVEYOR_BUILD_FOLDER%\build\VS2010\CompileAsCpp.props /t:Clean,Build /p:Platform=%PLATFORM% /p:Configuration=%CONFIGURATION% /logger:"C:\Program Files\AppVeyor\BuildAgent\Appveyor.MSBuildLogger.dll" &&
-      DIR build\VS2010\bin\%PLATFORM%_%CONFIGURATION%\*.exe &&
-      msbuild "build\VS2010\zstd.sln" /m /verbosity:minimal /property:PlatformToolset=v110 /t:Clean,Build /p:Platform=%PLATFORM% /p:Configuration=%CONFIGURATION% /logger:"C:\Program Files\AppVeyor\BuildAgent\Appveyor.MSBuildLogger.dll" &&
-      DIR build\VS2010\bin\%PLATFORM%_%CONFIGURATION%\*.exe
-    )
-
-
-  test_script:
-  - ECHO Testing %COMPILER% %PLATFORM% %CONFIGURATION%
-  - SET "FUZZERTEST=-T10s"
-  - if [%HOST%]==[mingw] (
-      set "CC=%COMPILER%" &&
-      make clean &&
-      make check
-    )

build/VS2008/zstd/zstd.vcproj

@@ -356,6 +356,10 @@
 				RelativePath="..\..\..\programs\dibio.c"
 				>
 			</File>
+			<File
+				RelativePath="..\..\..\programs\lorem.c"
+				>
+			</File>
 			<File
 				RelativePath="..\..\..\lib\dictBuilder\cover.c"
 				>

build/VS2010/datagen/datagen.vcxproj

@@ -157,6 +157,8 @@
   <ItemGroup>
     <ClCompile Include="..\..\..\programs\util.c" />
     <ClCompile Include="..\..\..\programs\datagen.c" />
+    <ClCompile Include="..\..\..\programs\lorem.c" />
+    <ClCompile Include="..\..\..\tests\loremOut.c" />
     <ClCompile Include="..\..\..\tests\datagencli.c" />
   </ItemGroup>
   <ItemGroup>

build/VS2010/libzstd-dll/libzstd-dll.vcxproj

@@ -169,7 +169,6 @@
       <WarningLevel>Level4</WarningLevel>
       <Optimization>Disabled</Optimization>
       <PreprocessorDefinitions>ZSTD_DLL_EXPORT=1;ZSTD_MULTITHREAD=1;ZSTD_LEGACY_SUPPORT=5;WIN32;_DEBUG;_CONSOLE;_CRT_SECURE_NO_WARNINGS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
-      <MinimalRebuild>true</MinimalRebuild>
       <BasicRuntimeChecks>EnableFastChecks</BasicRuntimeChecks>
       <RuntimeLibrary>MultiThreadedDebugDLL</RuntimeLibrary>
       <DebugInformationFormat>EditAndContinue</DebugInformationFormat>

build/VS2010/libzstd/libzstd.vcxproj

@@ -162,7 +162,6 @@
       <WarningLevel>Level4</WarningLevel>
       <Optimization>Disabled</Optimization>
       <PreprocessorDefinitions>ZSTD_MULTITHREAD=1;ZSTD_LEGACY_SUPPORT=5;WIN32;_DEBUG;_CONSOLE;_CRT_SECURE_NO_WARNINGS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
-      <MinimalRebuild>true</MinimalRebuild>
       <BasicRuntimeChecks>EnableFastChecks</BasicRuntimeChecks>
       <RuntimeLibrary>MultiThreadedDebugDLL</RuntimeLibrary>
       <DebugInformationFormat>EditAndContinue</DebugInformationFormat>

build/VS2010/zstd/zstd.vcxproj

@@ -63,6 +63,7 @@
     <ClCompile Include="..\..\..\programs\dibio.c" />
     <ClCompile Include="..\..\..\programs\fileio.c" />
     <ClCompile Include="..\..\..\programs\fileio_asyncio.c" />
+    <ClCompile Include="..\..\..\programs\lorem.c" />
     <ClCompile Include="..\..\..\programs\zstdcli.c" />
     <ClCompile Include="..\..\..\programs\zstdcli_trace.c" />
   </ItemGroup>
@@ -114,6 +115,7 @@
     <RootNamespace>zstd</RootNamespace>
     <OutDir>$(SolutionDir)bin\$(Platform)_$(Configuration)\</OutDir>
     <IntDir>$(SolutionDir)bin\obj\$(RootNamespace)_$(Platform)_$(Configuration)\</IntDir>
+    <InstructionSet>NotSet</InstructionSet>
   </PropertyGroup>
   <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="Configuration">
@@ -187,6 +189,7 @@
       <PreprocessorDefinitions>ZSTD_MULTITHREAD=1;ZSTD_LEGACY_SUPPORT=5;WIN32;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
       <TreatWarningAsError>true</TreatWarningAsError>
       <EnablePREfast>false</EnablePREfast>
+      <EnableEnhancedInstructionSet>$(InstructionSet)</EnableEnhancedInstructionSet>
     </ClCompile>
     <Link>
       <SubSystem>Console</SubSystem>
@@ -203,6 +206,7 @@
       <PreprocessorDefinitions>ZSTD_MULTITHREAD=1;ZSTD_LEGACY_SUPPORT=5;WIN32;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
       <TreatWarningAsError>true</TreatWarningAsError>
       <EnablePREfast>false</EnablePREfast>
+      <EnableEnhancedInstructionSet>$(InstructionSet)</EnableEnhancedInstructionSet>
     </ClCompile>
     <Link>
       <SubSystem>Console</SubSystem>
@@ -222,6 +226,7 @@
       <EnablePREfast>false</EnablePREfast>
       <TreatWarningAsError>false</TreatWarningAsError>
       <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+      <EnableEnhancedInstructionSet>$(InstructionSet)</EnableEnhancedInstructionSet>
     </ClCompile>
     <Link>
       <SubSystem>Console</SubSystem>
@@ -244,6 +249,7 @@
       <EnablePREfast>false</EnablePREfast>
       <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
       <AdditionalOptions>/DZSTD_MULTITHREAD %(AdditionalOptions)</AdditionalOptions>
+      <EnableEnhancedInstructionSet>$(InstructionSet)</EnableEnhancedInstructionSet>
     </ClCompile>
     <Link>
       <SubSystem>Console</SubSystem>

build/cmake/CMakeLists.txt

@@ -7,16 +7,14 @@
 # in the COPYING file in the root directory of this source tree).
 # ################################################################
 
-cmake_minimum_required(VERSION 2.8.12 FATAL_ERROR)
+cmake_minimum_required(VERSION 3.5 FATAL_ERROR)
 
 # As of 2018-12-26 ZSTD has been validated to build with cmake version 3.13.2 new policies.
 # Set and use the newest cmake policies that are validated to work
 set(ZSTD_MAX_VALIDATED_CMAKE_MAJOR_VERSION "3")
 set(ZSTD_MAX_VALIDATED_CMAKE_MINOR_VERSION "13") #Policies never changed at PATCH level
-if("${CMAKE_MAJOR_VERSION}" LESS 3)
-  set(ZSTD_CMAKE_POLICY_VERSION "${CMAKE_VERSION}")
-elseif( "${ZSTD_MAX_VALIDATED_CMAKE_MAJOR_VERSION}" EQUAL "${CMAKE_MAJOR_VERSION}" AND
-        "${ZSTD_MAX_VALIDATED_CMAKE_MINOR_VERSION}" GREATER "${CMAKE_MINOR_VERSION}")
+if("${ZSTD_MAX_VALIDATED_CMAKE_MAJOR_VERSION}" EQUAL "${CMAKE_MAJOR_VERSION}" AND
+       "${ZSTD_MAX_VALIDATED_CMAKE_MINOR_VERSION}" GREATER "${CMAKE_MINOR_VERSION}")
     set(ZSTD_CMAKE_POLICY_VERSION "${CMAKE_VERSION}")
 else()
     set(ZSTD_CMAKE_POLICY_VERSION "${ZSTD_MAX_VALIDATED_CMAKE_MAJOR_VERSION}.${ZSTD_MAX_VALIDATED_CMAKE_MINOR_VERSION}.0")
@@ -32,24 +30,13 @@ set(LIBRARY_DIR ${ZSTD_SOURCE_DIR}/lib)
 include(GetZstdLibraryVersion)
 GetZstdLibraryVersion(${LIBRARY_DIR}/zstd.h zstd_VERSION_MAJOR zstd_VERSION_MINOR zstd_VERSION_PATCH)
 
-if( CMAKE_MAJOR_VERSION LESS 3 )
-  ## Provide cmake 3+ behavior for older versions of cmake
-  project(zstd)
-  set(PROJECT_VERSION_MAJOR ${zstd_VERSION_MAJOR})
-  set(PROJECT_VERSION_MINOR ${zstd_VERSION_MINOR})
-  set(PROJECT_VERSION_PATCH ${zstd_VERSION_PATCH})
-  set(PROJECT_VERSION "${zstd_VERSION_MAJOR}.${zstd_VERSION_MINOR}.${zstd_VERSION_PATCH}")
-  enable_language(C)   # Main library is in C
-  enable_language(ASM) # And ASM
-  enable_language(CXX) # Testing contributed code also utilizes CXX
-else()
-  project(zstd
-    VERSION "${zstd_VERSION_MAJOR}.${zstd_VERSION_MINOR}.${zstd_VERSION_PATCH}"
-    LANGUAGES C   # Main library is in C
-              ASM # And ASM
-              CXX # Testing contributed code also utilizes CXX
-    )
-endif()
+project(zstd
+  VERSION "${zstd_VERSION_MAJOR}.${zstd_VERSION_MINOR}.${zstd_VERSION_PATCH}"
+  LANGUAGES C   # Main library is in C
+            ASM # And ASM
+            CXX # Testing contributed code also utilizes CXX
+  )
+
 message(STATUS "ZSTD VERSION: ${zstd_VERSION}")
 set(zstd_HOMEPAGE_URL "https://facebook.github.io/zstd")
 set(zstd_DESCRIPTION  "Zstandard is a real-time compression algorithm, providing high compression ratios.")
@@ -127,10 +114,26 @@ endif ()
 #-----------------------------------------------------------------------------
 # External dependencies
 #-----------------------------------------------------------------------------
+# Define a function to handle special thread settings for HP-UX
+# See https://github.com/facebook/zstd/pull/3862 for details.
+function(setup_hpux_threads)
+    find_package(Threads)
+    if (NOT Threads_FOUND)
+        set(CMAKE_USE_PTHREADS_INIT 1 PARENT_SCOPE)
+        set(CMAKE_THREAD_LIBS_INIT -lpthread PARENT_SCOPE)
+        set(CMAKE_HAVE_THREADS_LIBRARY 1 PARENT_SCOPE)
+        set(Threads_FOUND TRUE PARENT_SCOPE)
+    endif()
+endfunction()
+
 if (ZSTD_MULTITHREAD_SUPPORT AND UNIX)
-    set(THREADS_PREFER_PTHREAD_FLAG ON)
-    find_package(Threads REQUIRED)
-    if(CMAKE_USE_PTHREADS_INIT)
+    if (CMAKE_SYSTEM_NAME MATCHES "HP-UX")
+        setup_hpux_threads()
+    else()
+        set(THREADS_PREFER_PTHREAD_FLAG ON)
+        find_package(Threads REQUIRED)
+    endif()
+    if (CMAKE_USE_PTHREADS_INIT)
         set(THREADS_LIBS "${CMAKE_THREAD_LIBS_INIT}")
     else()
         message(SEND_ERROR "ZSTD currently does not support thread libraries other than pthreads")
@@ -193,10 +196,6 @@ export(EXPORT zstdExports
     FILE "${CMAKE_CURRENT_BINARY_DIR}/zstdTargets.cmake"
     NAMESPACE zstd::
     )
-configure_file(zstdConfig.cmake
-    "${CMAKE_CURRENT_BINARY_DIR}/zstdConfig.cmake"
-    COPYONLY
-    )
 
 # A Package Config file that works from the installation directory
 set(ConfigPackageLocation ${CMAKE_INSTALL_LIBDIR}/cmake/zstd)
@@ -205,8 +204,13 @@ install(EXPORT zstdExports
     NAMESPACE zstd::
     DESTINATION ${ConfigPackageLocation}
     )
+configure_package_config_file(
+    zstdConfig.cmake.in
+    "${CMAKE_CURRENT_BINARY_DIR}/zstdConfig.cmake"
+    INSTALL_DESTINATION ${ConfigPackageLocation}
+)
 install(FILES
-    zstdConfig.cmake
+    "${CMAKE_CURRENT_BINARY_DIR}/zstdConfig.cmake"
     "${CMAKE_CURRENT_BINARY_DIR}/zstdConfigVersion.cmake"
     DESTINATION ${ConfigPackageLocation}
     )

build/cmake/CMakeModules/AddZstdCompilationFlags.cmake

@@ -1,6 +1,15 @@
 include(CheckCXXCompilerFlag)
 include(CheckCCompilerFlag)
-include(CheckLinkerFlag)
+# VERSION_GREATER_EQUAL requires CMake 3.7 or later.
+# https://cmake.org/cmake/help/latest/command/if.html#version-greater-equal
+if (CMAKE_VERSION VERSION_LESS 3.18)
+    set(ZSTD_HAVE_CHECK_LINKER_FLAG false)
+else ()
+    set(ZSTD_HAVE_CHECK_LINKER_FLAG true)
+endif ()
+if (ZSTD_HAVE_CHECK_LINKER_FLAG)
+    include(CheckLinkerFlag)
+endif()
 
 function(EnableCompilerFlag _flag _C _CXX _LD)
     string(REGEX REPLACE "\\+" "PLUS" varname "${_flag}")
@@ -20,7 +29,20 @@ function(EnableCompilerFlag _flag _C _CXX _LD)
         endif ()
     endif ()
     if (_LD)
-        CHECK_LINKER_FLAG(C ${_flag} LD_FLAG_${varname})
+        # We never add a linker flag with CMake < 3.18. We will
+        # implement CHECK_LINKER_FLAG() like feature for CMake < 3.18
+        # or require CMake >= 3.18 when we need to add a required
+        # linker flag in future.
+        #
+        # We also skip linker flags check for MSVC compilers (which includes
+        # clang-cl) since currently check_linker_flag() doesn't give correct
+        # results for this configuration,
+        # see: https://gitlab.kitware.com/cmake/cmake/-/issues/22023
+        if (ZSTD_HAVE_CHECK_LINKER_FLAG AND NOT MSVC)
+            CHECK_LINKER_FLAG(C ${_flag} LD_FLAG_${varname})
+        else ()
+            set(LD_FLAG_${varname} false)
+        endif ()
         if (LD_FLAG_${varname})
             set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} ${_flag}" PARENT_SCOPE)
             set(CMAKE_SHARED_LINKER_FLAGS "${CMAKE_SHARED_LINKER_FLAGS} ${_flag}" PARENT_SCOPE)

build/cmake/README.md

@@ -41,6 +41,38 @@ cmake -DZSTD_BUILD_TESTS=ON -DZSTD_LEGACY_SUPPORT=OFF ..
 make
 ```
 
+### how to use it with CMake FetchContent
+
+For all options available, you can see it on <https://github.com/facebook/zstd/blob/dev/build/cmake/lib/CMakeLists.txt>
+```cmake
+include(FetchContent)
+
+set(ZSTD_BUILD_STATIC ON)
+set(ZSTD_BUILD_SHARED OFF)
+
+FetchContent_Declare(
+    zstd
+    URL "https://github.com/facebook/zstd/releases/download/v1.5.5/zstd-1.5.5.tar.gz"
+    DOWNLOAD_EXTRACT_TIMESTAMP TRUE
+    SOURCE_SUBDIR build/cmake
+)
+
+FetchContent_MakeAvailable(zstd)
+
+target_link_libraries(
+    ${PROJECT_NAME}
+    PRIVATE
+    libzstd_static
+)
+
+# On windows and macos this is needed
+target_include_directories(
+    ${PROJECT_NAME}
+    PRIVATE
+    ${zstd_SOURCE_DIR}/lib
+)
+```
+
 ### referring
 [Looking for a 'cmake clean' command to clear up CMake output](https://stackoverflow.com/questions/9680420/looking-for-a-cmake-clean-command-to-clear-up-cmake-output)
 

build/cmake/contrib/pzstd/CMakeLists.txt

@@ -18,6 +18,7 @@ set(PZSTD_DIR ${ZSTD_SOURCE_DIR}/contrib/pzstd)
 include_directories(${PROGRAMS_DIR} ${LIBRARY_DIR} ${LIBRARY_DIR}/common ${PZSTD_DIR})
 
 add_executable(pzstd ${PROGRAMS_DIR}/util.c ${PZSTD_DIR}/main.cpp ${PZSTD_DIR}/Options.cpp ${PZSTD_DIR}/Pzstd.cpp ${PZSTD_DIR}/SkippableFrame.cpp)
+target_compile_features(pzstd PRIVATE cxx_std_11)
 set_property(TARGET pzstd APPEND PROPERTY COMPILE_DEFINITIONS "NDEBUG")
 set_property(TARGET pzstd APPEND PROPERTY COMPILE_OPTIONS "-Wno-shadow")
 

build/cmake/lib/CMakeLists.txt

@@ -12,45 +12,70 @@ project(libzstd C ASM)
 set(CMAKE_INCLUDE_CURRENT_DIR TRUE)
 option(ZSTD_BUILD_STATIC "BUILD STATIC LIBRARIES" ON)
 option(ZSTD_BUILD_SHARED "BUILD SHARED LIBRARIES" ON)
+option(ZSTD_BUILD_COMPRESSION "BUILD COMPRESSION MODULE" ON)
+option(ZSTD_BUILD_DECOMPRESSION "BUILD DECOMPRESSION MODULE" ON)
+option(ZSTD_BUILD_DICTBUILDER "BUILD DICTBUILDER MODULE" ON)
+option(ZSTD_BUILD_DEPRECATED "BUILD DEPRECATED MODULE" OFF)
+
+set(ZSTDLIB_VISIBLE "" CACHE STRING "Visibility for ZSTDLIB API")
+set(ZSTDERRORLIB_VISIBLE "" CACHE STRING "Visibility for ZSTDERRORLIB_VISIBLE API")
+set(ZDICTLIB_VISIBLE "" CACHE STRING "Visibility for ZDICTLIB_VISIBLE API")
+set(ZSTDLIB_STATIC_API "" CACHE STRING "Visibility for ZSTDLIB_STATIC_API API")
+set(ZDICTLIB_STATIC_API "" CACHE STRING "Visibility for ZDICTLIB_STATIC_API API")
+
+set_property(CACHE ZSTDLIB_VISIBLE PROPERTY STRINGS "" "hidden" "default" "protected" "internal")
+set_property(CACHE ZSTDERRORLIB_VISIBLE PROPERTY STRINGS "" "hidden" "default" "protected" "internal")
+set_property(CACHE ZDICTLIB_VISIBLE PROPERTY STRINGS "" "hidden" "default" "protected" "internal")
+set_property(CACHE ZSTDLIB_STATIC_API PROPERTY STRINGS "" "hidden" "default" "protected" "internal")
+set_property(CACHE ZDICTLIB_STATIC_API PROPERTY STRINGS "" "hidden" "default" "protected" "internal")
 
 if(NOT ZSTD_BUILD_SHARED AND NOT ZSTD_BUILD_STATIC)
     message(SEND_ERROR "You need to build at least one flavor of libzstd")
 endif()
 
-# Define library directory, where sources and header files are located
-include_directories(${LIBRARY_DIR} ${LIBRARY_DIR}/common)
-
 file(GLOB CommonSources ${LIBRARY_DIR}/common/*.c)
 file(GLOB CompressSources ${LIBRARY_DIR}/compress/*.c)
+file(GLOB DecompressSources ${LIBRARY_DIR}/decompress/*.c)
 if (MSVC)
-    file(GLOB DecompressSources ${LIBRARY_DIR}/decompress/*.c)
     add_compile_options(-DZSTD_DISABLE_ASM)
 else ()
-    file(GLOB DecompressSources ${LIBRARY_DIR}/decompress/*.c ${LIBRARY_DIR}/decompress/*.S)
+    if(CMAKE_SYSTEM_PROCESSOR MATCHES "amd64.*|AMD64.*|x86_64.*|X86_64.*")
+        set(DecompressSources ${DecompressSources} ${LIBRARY_DIR}/decompress/huf_decompress_amd64.S)
+    else()
+        add_compile_options(-DZSTD_DISABLE_ASM)
+    endif()
 endif ()
 file(GLOB DictBuilderSources ${LIBRARY_DIR}/dictBuilder/*.c)
+file(GLOB DeprecatedSources ${LIBRARY_DIR}/deprecated/*.c)
 
-set(Sources
-        ${CommonSources}
-        ${CompressSources}
-        ${DecompressSources}
-        ${DictBuilderSources})
-
+file(GLOB PublicHeaders ${LIBRARY_DIR}/*.h)
 file(GLOB CommonHeaders ${LIBRARY_DIR}/common/*.h)
 file(GLOB CompressHeaders ${LIBRARY_DIR}/compress/*.h)
 file(GLOB DecompressHeaders ${LIBRARY_DIR}/decompress/*.h)
 file(GLOB DictBuilderHeaders ${LIBRARY_DIR}/dictBuilder/*.h)
+file(GLOB DeprecatedHeaders ${LIBRARY_DIR}/deprecated/*.h)
 
-set(Headers
-        ${LIBRARY_DIR}/zstd.h
-        ${CommonHeaders}
-        ${CompressHeaders}
-        ${DecompressHeaders}
-        ${DictBuilderHeaders})
+set(Sources ${CommonSources})
+set(Headers ${PublicHeaders} ${CommonHeaders})
+if (ZSTD_BUILD_COMPRESSION)
+    set(Sources ${Sources} ${CompressSources})
+    set(Headers ${Headers} ${CompressHeaders})
+endif()
+if (ZSTD_BUILD_DECOMPRESSION)
+    set(Sources ${Sources} ${DecompressSources})
+    set(Headers ${Headers} ${DecompressHeaders})
+endif()
+if (ZSTD_BUILD_DICTBUILDER)
+    set(Sources ${Sources} ${DictBuilderSources})
+    set(Headers ${Headers} ${DictBuilderHeaders})
+endif()
+if (ZSTD_BUILD_DEPRECATED)
+    set(Sources ${Sources} ${DeprecatedSources})
+    set(Headers ${Headers} ${DeprecatedHeaders})
+endif()
 
 if (ZSTD_LEGACY_SUPPORT)
     set(LIBRARY_LEGACY_DIR ${LIBRARY_DIR}/legacy)
-    include_directories(${LIBRARY_LEGACY_DIR})
 
     set(Sources ${Sources}
             ${LIBRARY_LEGACY_DIR}/zstd_v01.c
@@ -81,22 +106,38 @@ endif ()
 # Our assembly expects to be compiled by a C compiler, and is only enabled for
 # __GNUC__ compatible compilers. Otherwise all the ASM code is disabled by
 # macros.
-set_source_files_properties(${Sources} PROPERTIES LANGUAGE C)
+if(NOT CMAKE_ASM_COMPILER STREQUAL CMAKE_C_COMPILER)
+    set_source_files_properties(${Sources} PROPERTIES LANGUAGE C)
+endif()
+
+macro (add_definition target var)
+    if (NOT ("${${var}}" STREQUAL ""))
+        set_property(TARGET ${target} APPEND PROPERTY COMPILE_DEFINITIONS "${var}=__attribute__((visibility(\"${${var}}\")))")
+    endif ()
+endmacro ()
+
+# Define directories containing the library's public headers
+set(PUBLIC_INCLUDE_DIRS ${LIBRARY_DIR})
 
 # Split project to static and shared libraries build
 set(library_targets)
 if (ZSTD_BUILD_SHARED)
     add_library(libzstd_shared SHARED ${Sources} ${Headers} ${PlatformDependResources})
+    target_include_directories(libzstd_shared INTERFACE $<BUILD_INTERFACE:${PUBLIC_INCLUDE_DIRS}>)
     list(APPEND library_targets libzstd_shared)
     if (ZSTD_MULTITHREAD_SUPPORT)
         set_property(TARGET libzstd_shared APPEND PROPERTY COMPILE_DEFINITIONS "ZSTD_MULTITHREAD")
         if (UNIX)
             target_link_libraries(libzstd_shared ${THREADS_LIBS})
         endif ()
-    endif()
+    endif ()
+    add_definition(libzstd_shared ZSTDLIB_VISIBLE)
+    add_definition(libzstd_shared ZSTDERRORLIB_VISIBLE)
+    add_definition(libzstd_shared ZDICTLIB_VISIBLE)
 endif ()
 if (ZSTD_BUILD_STATIC)
     add_library(libzstd_static STATIC ${Sources} ${Headers})
+    target_include_directories(libzstd_static INTERFACE $<BUILD_INTERFACE:${PUBLIC_INCLUDE_DIRS}>)
     list(APPEND library_targets libzstd_static)
     if (ZSTD_MULTITHREAD_SUPPORT)
         set_property(TARGET libzstd_static APPEND PROPERTY COMPILE_DEFINITIONS "ZSTD_MULTITHREAD")
@@ -104,6 +145,41 @@ if (ZSTD_BUILD_STATIC)
             target_link_libraries(libzstd_static ${THREADS_LIBS})
         endif ()
     endif ()
+    add_definition(libzstd_static ZSTDLIB_VISIBLE)
+    add_definition(libzstd_static ZSTDERRORLIB_VISIBLE)
+    add_definition(libzstd_static ZDICTLIB_VISIBLE)
+    add_definition(libzstd_static ZSTDLIB_STATIC_API)
+    add_definition(libzstd_static ZDICTLIB_STATIC_API)
+endif ()
+if (ZSTD_BUILD_SHARED AND NOT ZSTD_BUILD_STATIC)
+    if (NOT BUILD_SHARED_LIBS)
+        message(WARNING "BUILD_SHARED_LIBS is OFF, but ZSTD_BUILD_SHARED is ON and ZSTD_BUILD_STATIC is OFF, which takes precedence, so libzstd is a shared library")
+    endif ()
+    add_library(libzstd INTERFACE)
+    target_link_libraries(libzstd INTERFACE libzstd_shared)
+    list(APPEND library_targets libzstd)
+endif ()
+if (ZSTD_BUILD_STATIC AND NOT ZSTD_BUILD_SHARED)
+    if (BUILD_SHARED_LIBS)
+        message(WARNING "BUILD_SHARED_LIBS is ON, but ZSTD_BUILD_SHARED is OFF and ZSTD_BUILD_STATIC is ON, which takes precedence, is set so libzstd is a static library")
+    endif ()
+    add_library(libzstd INTERFACE)
+    target_link_libraries(libzstd INTERFACE libzstd_static)
+    list(APPEND library_targets libzstd)
+endif ()
+if (ZSTD_BUILD_SHARED AND ZSTD_BUILD_STATIC)
+    # If both ZSTD_BUILD_SHARED and ZSTD_BUILD_STATIC are set, which is the
+    # default, fallback to using BUILD_SHARED_LIBS to determine whether to
+    # set libzstd to static or shared.
+    if (BUILD_SHARED_LIBS)
+        add_library(libzstd INTERFACE)
+        target_link_libraries(libzstd INTERFACE libzstd_shared)
+        list(APPEND library_targets libzstd)
+    else ()
+        add_library(libzstd INTERFACE)
+        target_link_libraries(libzstd INTERFACE libzstd_static)
+        list(APPEND library_targets libzstd)
+    endif ()
 endif ()
 
 # Add specific compile definitions for MSVC project
@@ -154,11 +230,7 @@ configure_file("${LIBRARY_DIR}/libzstd.pc.in" "${CMAKE_CURRENT_BINARY_DIR}/libzs
 install(FILES "${CMAKE_CURRENT_BINARY_DIR}/libzstd.pc" DESTINATION "${CMAKE_INSTALL_LIBDIR}/pkgconfig")
 
 # install target
-install(FILES
-    "${LIBRARY_DIR}/zstd.h"
-    "${LIBRARY_DIR}/zdict.h"
-    "${LIBRARY_DIR}/zstd_errors.h"
-    DESTINATION "${CMAKE_INSTALL_INCLUDEDIR}")
+install(FILES ${PublicHeaders} DESTINATION "${CMAKE_INSTALL_INCLUDEDIR}")
 
 install(TARGETS ${library_targets}
     EXPORT zstdExports

build/cmake/programs/CMakeLists.txt

@@ -32,7 +32,12 @@ if (MSVC)
     set(PlatformDependResources ${MSVC_RESOURCE_DIR}/zstd.rc)
 endif ()
 
-add_executable(zstd ${PROGRAMS_DIR}/zstdcli.c ${PROGRAMS_DIR}/util.c ${PROGRAMS_DIR}/timefn.c ${PROGRAMS_DIR}/fileio.c ${PROGRAMS_DIR}/fileio_asyncio.c ${PROGRAMS_DIR}/benchfn.c ${PROGRAMS_DIR}/benchzstd.c ${PROGRAMS_DIR}/datagen.c ${PROGRAMS_DIR}/dibio.c ${PROGRAMS_DIR}/zstdcli_trace.c ${PlatformDependResources})
+file(GLOB ZSTD_PROGRAM_SRCS "${PROGRAMS_DIR}/*.c")
+if (MSVC AND ZSTD_PROGRAMS_LINK_SHARED)
+    list(APPEND ZSTD_PROGRAM_SRCS ${LIBRARY_DIR}/common/pool.c ${LIBRARY_DIR}/common/threading.c)
+endif ()
+
+add_executable(zstd ${ZSTD_PROGRAM_SRCS})
 target_link_libraries(zstd ${PROGRAMS_ZSTD_LINK_TARGET})
 if (CMAKE_SYSTEM_NAME MATCHES "(Solaris|SunOS)")
     target_link_libraries(zstd rt)
@@ -75,7 +80,9 @@ if (UNIX)
         ${CMAKE_CURRENT_BINARY_DIR}/zstdless.1
         DESTINATION "${MAN_INSTALL_DIR}")
 
-    add_executable(zstd-frugal ${PROGRAMS_DIR}/zstdcli.c ${PROGRAMS_DIR}/util.c ${PROGRAMS_DIR}/timefn.c ${PROGRAMS_DIR}/fileio.c ${PROGRAMS_DIR}/fileio_asyncio.c)
+    add_executable(zstd-frugal ${PROGRAMS_DIR}/zstdcli.c
+        ${PROGRAMS_DIR}/util.c ${PROGRAMS_DIR}/timefn.c
+        ${PROGRAMS_DIR}/fileio.c ${PROGRAMS_DIR}/fileio_asyncio.c)
     target_link_libraries(zstd-frugal ${PROGRAMS_ZSTD_LINK_TARGET})
     set_property(TARGET zstd-frugal APPEND PROPERTY COMPILE_DEFINITIONS "ZSTD_NOBENCH;ZSTD_NODICT;ZSTD_NOTRACE")
 endif ()

build/cmake/tests/CMakeLists.txt

@@ -50,18 +50,18 @@ set(PROGRAMS_DIR ${ZSTD_SOURCE_DIR}/programs)
 set(TESTS_DIR ${ZSTD_SOURCE_DIR}/tests)
 include_directories(${TESTS_DIR} ${PROGRAMS_DIR} ${LIBRARY_DIR} ${LIBRARY_DIR}/common ${LIBRARY_DIR}/compress ${LIBRARY_DIR}/dictBuilder)
 
-add_executable(datagen ${PROGRAMS_DIR}/datagen.c ${TESTS_DIR}/datagencli.c)
+add_executable(datagen ${PROGRAMS_DIR}/datagen.c ${PROGRAMS_DIR}/lorem.c ${TESTS_DIR}/loremOut.c ${TESTS_DIR}/datagencli.c)
 target_link_libraries(datagen libzstd_static)
 
 #
 # fullbench
 #
-add_executable(fullbench ${PROGRAMS_DIR}/datagen.c ${PROGRAMS_DIR}/util.c ${PROGRAMS_DIR}/timefn.c ${PROGRAMS_DIR}/benchfn.c ${PROGRAMS_DIR}/benchzstd.c ${TESTS_DIR}/fullbench.c)
+add_executable(fullbench ${PROGRAMS_DIR}/datagen.c ${PROGRAMS_DIR}/lorem.c ${PROGRAMS_DIR}/util.c ${PROGRAMS_DIR}/timefn.c ${PROGRAMS_DIR}/benchfn.c ${PROGRAMS_DIR}/benchzstd.c ${TESTS_DIR}/fullbench.c)
 if (NOT MSVC)
     target_compile_options(fullbench PRIVATE "-Wno-deprecated-declarations")
 endif()
 target_link_libraries(fullbench libzstd_static)
-add_test(NAME fullbench COMMAND fullbench ${ZSTD_FULLBENCH_FLAGS})
+add_test(NAME fullbench COMMAND "$<TARGET_FILE:fullbench>" ${ZSTD_FULLBENCH_FLAGS})
 
 #
 # fuzzer
@@ -73,7 +73,7 @@ endif()
 target_link_libraries(fuzzer libzstd_static)
 AddTestFlagsOption(ZSTD_FUZZER_FLAGS "$ENV{FUZZERTEST} $ENV{FUZZER_FLAGS}"
     "Semicolon-separated list of flags to pass to the fuzzer test (see `fuzzer -h` for usage)")
-add_test(NAME fuzzer COMMAND fuzzer ${ZSTD_FUZZER_FLAGS})
+add_test(NAME fuzzer COMMAND "$<TARGET_FILE:fuzzer>" ${ZSTD_FUZZER_FLAGS})
 # Disable the timeout since the run time is too long for the default timeout of
 # 1500 seconds and varies considerably between low-end and high-end CPUs.
 # set_tests_properties(fuzzer PROPERTIES TIMEOUT 0)
@@ -88,7 +88,7 @@ endif()
 target_link_libraries(zstreamtest libzstd_static)
 AddTestFlagsOption(ZSTD_ZSTREAM_FLAGS "$ENV{ZSTREAM_TESTTIME} $ENV{FUZZER_FLAGS}"
     "Semicolon-separated list of flags to pass to the zstreamtest test (see `zstreamtest -h` for usage)")
-add_test(NAME zstreamtest COMMAND zstreamtest ${ZSTD_ZSTREAM_FLAGS})
+add_test(NAME zstreamtest COMMAND "$<TARGET_FILE:zstreamtest>" ${ZSTD_ZSTREAM_FLAGS})
 
 #
 # playTests.sh
@@ -110,7 +110,7 @@ endif()
 # Label the "Medium" set of tests (see TESTING.md)
 set_property(TEST fuzzer zstreamtest playTests APPEND PROPERTY LABELS Medium)
 
-add_executable(paramgrill ${PROGRAMS_DIR}/benchfn.c ${PROGRAMS_DIR}/benchzstd.c ${PROGRAMS_DIR}/datagen.c ${PROGRAMS_DIR}/util.c ${PROGRAMS_DIR}/timefn.c ${TESTS_DIR}/paramgrill.c)
+add_executable(paramgrill ${PROGRAMS_DIR}/benchfn.c ${PROGRAMS_DIR}/benchzstd.c ${PROGRAMS_DIR}/datagen.c ${PROGRAMS_DIR}/lorem.c ${PROGRAMS_DIR}/util.c ${PROGRAMS_DIR}/timefn.c ${TESTS_DIR}/paramgrill.c)
 if (UNIX)
     target_link_libraries(paramgrill libzstd_static m) #m is math library
 else()

build/cmake/zstdConfig.cmake

@@ -1 +0,0 @@
-include("${CMAKE_CURRENT_LIST_DIR}/zstdTargets.cmake")

build/cmake/zstdConfig.cmake.in

@@ -0,0 +1,10 @@
+@PACKAGE_INIT@
+
+include(CMakeFindDependencyMacro)
+if(@ZSTD_MULTITHREAD_SUPPORT@ AND "@UNIX@")
+  find_dependency(Threads)
+endif()
+
+include("${CMAKE_CURRENT_LIST_DIR}/zstdTargets.cmake")
+
+check_required_components("zstd")

build/meson/meson.build

@@ -132,7 +132,7 @@ endif
 
 subdir('lib')
 
-if bin_programs
+if bin_programs or bin_tests
   subdir('programs')
 endif
 

build/meson/programs/meson.build

@@ -18,6 +18,7 @@ zstd_programs_sources = [join_paths(zstd_rootdir, 'programs/zstdcli.c'),
   join_paths(zstd_rootdir, 'programs/benchfn.c'),
   join_paths(zstd_rootdir, 'programs/benchzstd.c'),
   join_paths(zstd_rootdir, 'programs/datagen.c'),
+  join_paths(zstd_rootdir, 'programs/lorem.c'),
   join_paths(zstd_rootdir, 'programs/dibio.c'),
   join_paths(zstd_rootdir, 'programs/zstdcli_trace.c')]
 
@@ -72,7 +73,14 @@ zstd = executable('zstd',
   c_args: zstd_c_args,
   dependencies: zstd_deps,
   export_dynamic: export_dynamic_on_windows, # Since Meson 0.45.0
-  install: true)
+  build_by_default: bin_programs,
+  install: bin_programs)
+
+if not bin_programs
+  # we generate rules to build the programs, but don't install anything
+  # so do not continue to installing scripts and manpages
+  subdir_done()
+endif
 
 zstd_frugal_sources = [join_paths(zstd_rootdir, 'programs/zstdcli.c'),
   join_paths(zstd_rootdir, 'programs/timefn.c'),

build/meson/tests/meson.build

@@ -29,6 +29,7 @@ DECODECORPUS_TESTTIME = '-T30'
 test_includes = [ include_directories(join_paths(zstd_rootdir, 'programs')) ]
 
 testcommon_sources = [join_paths(zstd_rootdir, 'programs/datagen.c'),
+  join_paths(zstd_rootdir, 'programs/lorem.c'),
   join_paths(zstd_rootdir, 'programs/util.c'),
   join_paths(zstd_rootdir, 'programs/timefn.c'),
   join_paths(zstd_rootdir, 'programs/benchfn.c'),
@@ -43,7 +44,8 @@ testcommon_dep = declare_dependency(link_with: testcommon,
   dependencies: libzstd_deps,
   include_directories: libzstd_includes)
 
-datagen_sources = [join_paths(zstd_rootdir, 'tests/datagencli.c')]
+datagen_sources = [join_paths(zstd_rootdir, 'tests/datagencli.c'),
+  join_paths(zstd_rootdir, 'tests/loremOut.c')]
 datagen = executable('datagen',
   datagen_sources,
   c_args: [ '-DNDEBUG' ],
@@ -162,7 +164,7 @@ if host_machine_os != os_windows
       playTests_sh,
       args: opt,
       env: ['ZSTD_BIN=' + zstd.full_path(), 'DATAGEN_BIN=./datagen'],
-      depends: [datagen],
+      depends: [datagen, zstd],
       suite: suite,
       workdir: meson.current_build_dir(),
       timeout: 2800) # Timeout should work on HDD drive

contrib/docker/Dockerfile

@@ -1,13 +1,13 @@
 # Dockerfile
 # First image to build the binary
-FROM alpine as builder
+FROM alpine@sha256:69665d02cb32192e52e07644d76bc6f25abeb5410edc1c7a81a10ba3f0efb90a as builder
 
 RUN apk --no-cache add make gcc libc-dev
 COPY . /src
 RUN mkdir /pkg && cd /src && make && make DESTDIR=/pkg install
 
 # Second minimal image to only keep the built binary
-FROM alpine
+FROM alpine@sha256:69665d02cb32192e52e07644d76bc6f25abeb5410edc1c7a81a10ba3f0efb90a
 
 # Copy the built files
 COPY --from=builder /pkg /

contrib/linux-kernel/mem.h

@@ -24,6 +24,7 @@
 /*-****************************************
 *  Compiler specifics
 ******************************************/
+#undef MEM_STATIC /* may be already defined from common/compiler.h */
 #define MEM_STATIC static inline
 
 /*-**************************************************************

contrib/linux-kernel/zstd_common_module.c

@@ -24,9 +24,6 @@ EXPORT_SYMBOL_GPL(HUF_readStats_wksp);
 EXPORT_SYMBOL_GPL(ZSTD_isError);
 EXPORT_SYMBOL_GPL(ZSTD_getErrorName);
 EXPORT_SYMBOL_GPL(ZSTD_getErrorCode);
-EXPORT_SYMBOL_GPL(ZSTD_customMalloc);
-EXPORT_SYMBOL_GPL(ZSTD_customCalloc);
-EXPORT_SYMBOL_GPL(ZSTD_customFree);
 
 MODULE_LICENSE("Dual BSD/GPL");
 MODULE_DESCRIPTION("Zstd Common");

contrib/linux-kernel/zstd_decompress_module.c

@@ -77,7 +77,7 @@ EXPORT_SYMBOL(zstd_init_dstream);
 
 size_t zstd_reset_dstream(zstd_dstream *dstream)
 {
-	return ZSTD_resetDStream(dstream);
+	return ZSTD_DCtx_reset(dstream, ZSTD_reset_session_only);
 }
 EXPORT_SYMBOL(zstd_reset_dstream);
 

contrib/linux-kernel/zstd_deps.h

@@ -84,7 +84,7 @@ static uint64_t ZSTD_div64(uint64_t dividend, uint32_t divisor) {
 
 #include <linux/kernel.h>
 
-#define assert(x) WARN_ON((x))
+#define assert(x) WARN_ON(!(x))
 
 #endif /* ZSTD_DEPS_ASSERT */
 #endif /* ZSTD_DEPS_NEED_ASSERT */
@@ -115,11 +115,7 @@ static uint64_t ZSTD_div64(uint64_t dividend, uint32_t divisor) {
 #ifndef ZSTD_DEPS_STDINT
 #define ZSTD_DEPS_STDINT
 
-/*
- * The Linux Kernel doesn't provide intptr_t, only uintptr_t, which
- * is an unsigned long.
- */
-typedef long intptr_t;
+/* intptr_t already provided by ZSTD_DEPS_COMMON */
 
 #endif /* ZSTD_DEPS_STDINT */
 #endif /* ZSTD_DEPS_NEED_STDINT */

contrib/pzstd/Makefile

@@ -10,7 +10,7 @@
 # Standard variables for installation
 DESTDIR ?=
 PREFIX  ?= /usr/local
-BINDIR  := $(DESTDIR)$(PREFIX)/bin
+BINDIR  := $(PREFIX)/bin
 
 ZSTDDIR = ../../lib
 PROGDIR = ../../programs
@@ -37,10 +37,13 @@ CFLAGS   += -Wno-deprecated-declarations
 PZSTD_INC  = -I$(ZSTDDIR) -I$(ZSTDDIR)/common -I$(PROGDIR) -I.
 GTEST_INC  = -isystem googletest/googletest/include
 
+# Set the minimum required by gtest
+PZSTD_CXX_STD := -std=c++14
+
 PZSTD_CPPFLAGS  = $(PZSTD_INC)
 PZSTD_CCXXFLAGS =
 PZSTD_CFLAGS    = $(PZSTD_CCXXFLAGS)
-PZSTD_CXXFLAGS  = $(PZSTD_CCXXFLAGS) -std=c++11
+PZSTD_CXXFLAGS  = $(PZSTD_CCXXFLAGS) $(PZSTD_CXX_STD)
 PZSTD_LDFLAGS   =
 EXTRA_FLAGS     =
 ALL_CFLAGS      = $(EXTRA_FLAGS) $(CPPFLAGS) $(PZSTD_CPPFLAGS) $(CFLAGS)   $(PZSTD_CFLAGS)
@@ -106,12 +109,12 @@ check:
 .PHONY: install
 install: PZSTD_CPPFLAGS += -DNDEBUG
 install: pzstd$(EXT)
-	install -d -m 755 $(BINDIR)/
-	install -m 755 pzstd$(EXT) $(BINDIR)/pzstd$(EXT)
+	install -d -m 755 $(DESTDIR)$(BINDIR)/
+	install -m 755 pzstd$(EXT) $(DESTDIR)$(BINDIR)/pzstd$(EXT)
 
 .PHONY: uninstall
 uninstall:
-	$(RM) $(BINDIR)/pzstd$(EXT)
+	$(RM) $(DESTDIR)$(BINDIR)/pzstd$(EXT)
 
 # Targets for many different builds
 .PHONY: all

contrib/seekable_format/README.md

@@ -0,0 +1,42 @@
+# Zstandard Seekable Format
+
+The seekable format splits compressed data into a series of independent "frames",
+each compressed individually,
+so that decompression of a section in the middle of an archive
+only requires zstd to decompress at most a frame's worth of extra data,
+instead of the entire archive.
+
+The frames are appended, so that the decompression of the entire payload
+still regenerates the original content, using any compliant zstd decoder.
+
+On top of that, the seekable format generates a jump table,
+which makes it possible to jump directly to the position of the relevant frame
+when requesting only a segment of the data.
+The jump table is simply ignored by zstd decoders unaware of the seekable format.
+
+The format is delivered with an API to create seekable archives
+and to retrieve arbitrary segments inside the archive.
+
+### Maximum Frame Size parameter
+
+When creating a seekable archive, the main parameter is the maximum frame size.
+
+At compression time, user can manually select the boundaries between segments,
+but they don't have to: long segments will be automatically split
+when larger than selected maximum frame size.
+
+Small frame sizes reduce decompression cost when requesting small segments,
+because the decoder will nonetheless have to decompress an entire frame
+to recover just a single byte from it.
+
+A good rule of thumb is to select a maximum frame size roughly equivalent
+to the access pattern when it's known.
+For example, if the application tends to request 4KB blocks,
+then it's a good idea to set a maximum frame size in the vicinity of 4 KB.
+
+But small frame sizes also reduce compression ratio,
+and increase the cost for the jump table,
+so there is a balance to find.
+
+In general, try to avoid really tiny frame sizes (<1 KB),
+which would have a large negative impact on compression ratio.

contrib/seekable_format/examples/parallel_compression.c

@@ -25,7 +25,7 @@
 
 #include "pool.h"      // use zstd thread pool for demo
 
-#include "zstd_seekable.h"
+#include "../zstd_seekable.h"
 
 static void* malloc_orDie(size_t size)
 {

contrib/seekable_format/examples/parallel_processing.c

@@ -19,7 +19,7 @@
 #define ZSTD_STATIC_LINKING_ONLY
 #include <zstd.h>      // presumes zstd library is installed
 #include <zstd_errors.h>
-#if defined(WIN32) || defined(_WIN32)
+#if defined(_WIN32)
 #  include <windows.h>
 #  define SLEEP(x) Sleep(x)
 #else
@@ -29,7 +29,7 @@
 
 #include "pool.h"      // use zstd thread pool for demo
 
-#include "zstd_seekable.h"
+#include "../zstd_seekable.h"
 
 #define MIN(a, b) ((a) < (b) ? (a) : (b))
 

contrib/seekable_format/examples/seekable_compression.c

@@ -13,7 +13,7 @@
 #define ZSTD_STATIC_LINKING_ONLY
 #include <zstd.h>      // presumes zstd library is installed
 
-#include "zstd_seekable.h"
+#include "../zstd_seekable.h"
 
 static void* malloc_orDie(size_t size)
 {
@@ -112,20 +112,23 @@ static char* createOutFilename_orDie(const char* filename)
     return (char*)outSpace;
 }
 
-int main(int argc, const char** argv) {
+#define CLEVEL_DEFAULT 5
+int main(int argc, const char** argv)
+{
     const char* const exeName = argv[0];
-    if (argc!=3) {
-        printf("wrong arguments\n");
-        printf("usage:\n");
-        printf("%s FILE FRAME_SIZE\n", exeName);
+    if (argc<3 || argc>4) {
+        printf("wrong arguments \n");
+        printf("usage: \n");
+        printf("%s FILE FRAME_SIZE [LEVEL] \n", exeName);
         return 1;
     }
 
     {   const char* const inFileName = argv[1];
         unsigned const frameSize = (unsigned)atoi(argv[2]);
+        int const cLevel = (argc==4) ? atoi(argv[3]) : CLEVEL_DEFAULT;
 
         char* const outFileName = createOutFilename_orDie(inFileName);
-        compressFile_orDie(inFileName, outFileName, 5, frameSize);
+        compressFile_orDie(inFileName, outFileName, cLevel, frameSize);
         free(outFileName);
     }
 

contrib/seekable_format/examples/seekable_decompression.c

@@ -16,7 +16,7 @@
 #include <zstd.h>      // presumes zstd library is installed
 #include <zstd_errors.h>
 
-#include "zstd_seekable.h"
+#include "../zstd_seekable.h"
 
 #define MIN(a, b) ((a) < (b) ? (a) : (b))
 

contrib/seekable_format/tests/seekable_tests.c

@@ -3,8 +3,55 @@
 #include <stdlib.h>  // malloc
 #include <stdio.h>
 #include <assert.h>
+#include <string.h>
 
-#include "zstd_seekable.h"
+#include "../zstd_seekable.h"
+
+
+/* ZSTD_seekable_customFile implementation that reads/seeks a buffer while keeping track of total bytes read */
+typedef struct {
+    const void *ptr;
+    size_t size;
+    size_t pos;
+    size_t totalRead;
+} buffWrapperWithTotal_t;
+
+static int readBuffWithTotal(void* opaque, void* buffer, size_t n)
+{
+    buffWrapperWithTotal_t* const buff = (buffWrapperWithTotal_t*)opaque;
+    assert(buff != NULL);
+    if (buff->pos + n > buff->size) return -1;
+    memcpy(buffer, (const char*)buff->ptr + buff->pos, n);
+    buff->pos += n;
+    buff->totalRead += n;
+    return 0;
+}
+
+static int seekBuffWithTotal(void* opaque, long long offset, int origin)
+{
+    buffWrapperWithTotal_t* const buff = (buffWrapperWithTotal_t*) opaque;
+    unsigned long long newOffset;
+    assert(buff != NULL);
+    switch (origin) {
+    case SEEK_SET:
+        assert(offset >= 0);
+        newOffset = (unsigned long long)offset;
+        break;
+    case SEEK_CUR:
+        newOffset = (unsigned long long)((long long)buff->pos + offset);
+        break;
+    case SEEK_END:
+        newOffset = (unsigned long long)((long long)buff->size + offset);
+        break;
+    default:
+        assert(0);  /* not possible */
+    }
+    if (newOffset > buff->size) {
+        return -1;
+    }
+    buff->pos = newOffset;
+    return 0;
+}
 
 /* Basic unit tests for zstd seekable format */
 int main(int argc, const char** argv)
@@ -220,6 +267,92 @@ int main(int argc, const char** argv)
     }
     printf("Success!\n");
 
+
+    printf("Test %u - multiple decompress calls: ", testNb++);
+    {   char const inBuffer[] = "Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt";
+        size_t const inSize = sizeof(inBuffer);
+
+        size_t const seekCapacity = 5000;
+        void* const seekBuffer = malloc(seekCapacity);
+        assert(seekBuffer != NULL);
+        size_t seekSize;
+
+        size_t const outCapacity = inSize;
+        char* const outBuffer = malloc(outCapacity);
+        assert(outBuffer != NULL);
+
+        ZSTD_seekable_CStream* const zscs = ZSTD_seekable_createCStream();
+        assert(zscs != NULL);
+
+        /* compress test data with a small frame size to ensure multiple frames in the output */
+        unsigned const maxFrameSize = 40;
+        { size_t const initStatus = ZSTD_seekable_initCStream(zscs, 9, 0 /* checksumFlag */, maxFrameSize);
+          assert(!ZSTD_isError(initStatus));
+        }
+
+        {   ZSTD_outBuffer outb = { .dst=seekBuffer, .pos=0, .size=seekCapacity };
+            ZSTD_inBuffer inb = { .src=inBuffer, .pos=0, .size=inSize };
+
+            while (inb.pos < inb.size) {
+                size_t const cStatus = ZSTD_seekable_compressStream(zscs, &outb, &inb);
+                assert(!ZSTD_isError(cStatus));
+            }
+
+            size_t const endStatus = ZSTD_seekable_endStream(zscs, &outb);
+            assert(!ZSTD_isError(endStatus));
+            seekSize = outb.pos;
+        }
+
+        ZSTD_seekable* const stream = ZSTD_seekable_create();
+        assert(stream != NULL);
+        buffWrapperWithTotal_t buffWrapper = {seekBuffer, seekSize, 0, 0};
+        { ZSTD_seekable_customFile srcFile = {&buffWrapper, &readBuffWithTotal, &seekBuffWithTotal};
+          size_t const initStatus = ZSTD_seekable_initAdvanced(stream, srcFile);
+          assert(!ZSTD_isError(initStatus)); }
+
+        /* Perform a series of small reads and seeks (repeatedly read 1 byte and skip 1 byte)
+           and check that we didn't reread input data unnecessarily */
+        size_t pos;
+        for (pos = 0; pos < inSize; pos += 2) {
+            size_t const decStatus = ZSTD_seekable_decompress(stream, outBuffer, 1, pos);
+            if (decStatus != 1 || outBuffer[0] != inBuffer[pos]) {
+                goto _test_error;
+            }
+        }
+        if (buffWrapper.totalRead > seekSize) {
+            /* We read more than the compressed size, meaning there were some rereads.
+               This is unneeded because we only seeked forward. */
+            printf("Too much data read: %zu read, with compressed size %zu\n", buffWrapper.totalRead, seekSize);
+            goto _test_error;
+        }
+
+        /* Perform some reads and seeks to ensure correctness */
+        struct {
+            size_t offset;
+            size_t size;
+        } const tests[] = {  /* Assume the frame size is 40 */
+            {20, 40}, /* read partial data from two frames */
+            {60, 10}, /* continue reading from the same offset */
+            {50, 20}, /* seek backward within the same frame */
+            {10, 10}, /* seek backward to a different frame */
+            {25, 10}, /* seek forward within the same frame */
+            {60, 10}, /* seek forward to a different frame */
+        };
+        size_t idx;
+        for (idx = 0; idx < sizeof(tests) / sizeof(tests[0]); idx++) {
+            size_t const decStatus = ZSTD_seekable_decompress(stream, outBuffer, tests[idx].size, tests[idx].offset);
+            if (decStatus != tests[idx].size || memcmp(outBuffer, inBuffer + tests[idx].offset, tests[idx].size) != 0) {
+                goto _test_error;
+            }
+        }
+
+        free(seekBuffer);
+        free(outBuffer);
+        ZSTD_seekable_freeCStream(zscs);
+        ZSTD_seekable_free(stream);
+    }
+    printf("Success!\n");
+
     /* TODO: Add more tests */
     printf("Finished tests\n");
     return 0;

contrib/seekable_format/zstd_seekable.h

@@ -15,8 +15,8 @@ extern "C" {
 
 #define ZSTD_SEEKABLE_MAXFRAMES 0x8000000U
 
-/* Limit the maximum size to avoid any potential issues storing the compressed size */
-#define ZSTD_SEEKABLE_MAX_FRAME_DECOMPRESSED_SIZE 0x80000000U
+/* Limit maximum size to avoid potential issues storing the compressed size */
+#define ZSTD_SEEKABLE_MAX_FRAME_DECOMPRESSED_SIZE 0x40000000U
 
 /*-****************************************************************************
 *  Seekable Format
@@ -48,10 +48,19 @@ typedef struct ZSTD_seekTable_s ZSTD_seekTable;
 *
 *  Use ZSTD_seekable_initCStream() to initialize a ZSTD_seekable_CStream object
 *  for a new compression operation.
-*  `maxFrameSize` indicates the size at which to automatically start a new
-*  seekable frame.  `maxFrameSize == 0` implies the default maximum size.
-*  `checksumFlag` indicates whether or not the seek table should include frame
-*  checksums on the uncompressed data for verification.
+*  - `maxFrameSize` indicates the size at which to automatically start a new
+*            seekable frame.
+*            `maxFrameSize == 0` implies the default maximum size.
+*            Smaller frame sizes allow faster decompression of small segments,
+*            since retrieving a single byte requires decompression of
+*            the full frame where the byte belongs.
+*            In general, size the frames to roughly correspond to
+*            the access granularity (when it's known).
+*            But small sizes also reduce compression ratio.
+*            Avoid really tiny frame sizes (< 1 KB),
+*            that would hurt compression ratio considerably.
+*  - `checksumFlag` indicates whether or not the seek table should include frame
+*            checksums on the uncompressed data for verification.
 *  @return : a size hint for input to provide for compression, or an error code
 *            checkable with ZSTD_isError()
 *

contrib/seekable_format/zstdseek_compress.c

@@ -230,6 +230,8 @@ size_t ZSTD_seekable_compressStream(ZSTD_seekable_CStream* zcs, ZSTD_outBuffer*
     const BYTE* const inBase = (const BYTE*) input->src + input->pos;
     size_t inLen = input->size - input->pos;
 
+    assert(zcs->maxFrameSize < INT_MAX);
+    ZSTD_CCtx_setParameter(zcs->cstream, ZSTD_c_srcSizeHint, (int)zcs->maxFrameSize);
     inLen = MIN(inLen, (size_t)(zcs->maxFrameSize - zcs->frameDSize));
 
     /* if we haven't finished flushing the last frame, don't start writing a new one */

contrib/seekable_format/zstdseek_decompress.c

@@ -493,7 +493,7 @@ size_t ZSTD_seekable_decompress(ZSTD_seekable* zs, void* dst, size_t len, unsign
     size_t srcBytesRead = 0;
     do {
         /* check if we can continue from a previous decompress job */
-        if (targetFrame != zs->curFrame || offset != zs->decompressedOffset) {
+        if (targetFrame != zs->curFrame || offset < zs->decompressedOffset) {
             zs->decompressedOffset = zs->seekTable.entries[targetFrame].dOffset;
             zs->curFrame = targetFrame;
 

doc/decompressor_errata.md

@@ -6,12 +6,53 @@ Each entry will contain:
 1. The last affected decompressor versions.
 2. The decompressor components affected.
 2. Whether the compressed frame could ever be produced by the reference compressor.
-3. An example frame.
+3. An example frame (hexadecimal string when it can be short enough, link to golden file otherwise)
 4. A description of the bug.
 
 The document is in reverse chronological order, with the bugs that affect the most recent zstd decompressor versions listed first.
 
 
+No sequence using the 2-bytes format
+------------------------------------------------
+
+**Last affected version**: v1.5.5
+
+**Affected decompressor component(s)**: Library & CLI
+
+**Produced by the reference compressor**: No
+
+**Example Frame**: see zstd/tests/golden-decompression/zeroSeq_2B.zst
+
+The zstd decoder incorrectly expects FSE tables when there are 0 sequences present in the block
+if the value 0 is encoded using the 2-bytes format.
+Instead, it should immediately end the sequence section, and move on to next block.
+
+This situation was never generated by the reference compressor,
+because representing 0 sequences with the 2-bytes format is inefficient
+(the 1-byte format is always used in this case).
+
+
+Compressed block with a size of exactly 128 KB
+------------------------------------------------
+
+**Last affected version**: v1.5.2
+
+**Affected decompressor component(s)**: Library & CLI
+
+**Produced by the reference compressor**: No
+
+**Example Frame**: see zstd/tests/golden-decompression/block-128k.zst
+
+The zstd decoder incorrectly rejected blocks of type `Compressed_Block` when their size was exactly 128 KB.
+Note that `128 KB - 1` was accepted, and `128 KB + 1` is forbidden by the spec.
+
+This type of block was never generated by the reference compressor.
+
+These blocks used to be disallowed by the spec up until spec version 0.3.2 when the restriction was lifted by [PR#1689](https://github.com/facebook/zstd/pull/1689).
+
+> A Compressed_Block has the extra restriction that Block_Size is always strictly less than the decompressed size. If this condition cannot be respected, the block must be sent uncompressed instead (Raw_Block).
+
+
 Compressed block with 0 literals and 0 sequences
 ------------------------------------------------
 
@@ -31,6 +72,7 @@ Additionally, these blocks were disallowed by the spec up until spec version 0.3
 
 > A Compressed_Block has the extra restriction that Block_Size is always strictly less than the decompressed size. If this condition cannot be respected, the block must be sent uncompressed instead (Raw_Block).
 
+
 First block is RLE block
 ------------------------
 
@@ -52,6 +94,7 @@ block.
 
 https://github.com/facebook/zstd/blob/8814aa5bfa74f05a86e55e9d508da177a893ceeb/lib/compress/zstd_compress.c#L3527-L3535
 
+
 Tiny FSE Table & Block
 ----------------------
 
@@ -82,3 +125,24 @@ The total `Block_Content` is `5` bytes, and `Last_Table_Offset` is `2`.
 See the compressor workaround code:
 
 https://github.com/facebook/zstd/blob/8814aa5bfa74f05a86e55e9d508da177a893ceeb/lib/compress/zstd_compress.c#L2667-L2682
+
+Magicless format
+----------------------
+
+**Last affected version**: v1.5.5
+
+**Affected decompressor component(s)**: Library
+
+**Produced by the reference compressor**: Yes (example: https://gist.github.com/embg/9940726094f4cf2cef162cffe9319232)
+
+**Example Frame**: `27 b5 2f fd 00 03 19 00 00 66 6f 6f 3f ba c4 59`
+
+v1.5.6 fixes several bugs in which the magicless-format decoder rejects valid frames.
+These include but are not limited to:
+* Valid frames that happen to begin with a legacy magic number (little-endian)
+* Valid frames that happen to begin with a skippable magic number (little-endian)
+
+If you are affected by this issue and cannot update to v1.5.6 or later, there is a
+workaround to recover affected data. Simply prepend the ZSTD magic number
+`0xFD2FB528` (little-endian) to your data and decompress using the standard-format
+decoder.

doc/decompressor_permissive.md

@@ -0,0 +1,60 @@
+Decompressor Permissiveness to Invalid Data
+===========================================
+
+This document describes the behavior of the reference decompressor in cases
+where it accepts formally invalid data instead of reporting an error.
+
+While the reference decompressor *must* decode any compliant frame following
+the specification, its ability to detect erroneous data is on a best effort
+basis: the decoder may accept input data that would be formally invalid,
+when it causes no risk to the decoder, and which detection would cost too much
+complexity or speed regression.
+
+In practice, the vast majority of invalid data are detected, if only because
+many corruption events are dangerous for the decoder process (such as
+requesting an out-of-bound memory access) and many more are easy to check.
+
+This document lists a few known cases where invalid data was formerly accepted
+by the decoder, and what has changed since.
+
+
+Offset == 0
+-----------
+
+**Last affected version**: v1.5.5
+
+**Produced by the reference compressor**: No
+
+**Example Frame**: `28b5 2ffd 0000 4500 0008 0002 002f 430b ae`
+
+If a sequence is decoded with `literals_length = 0` and `offset_value = 3`
+while `Repeated_Offset_1 = 1`, the computed offset will be `0`, which is
+invalid.
+
+The reference decompressor up to v1.5.5 processes this case as if the computed
+offset was `1`, including inserting `1` into the repeated offset list.
+This prevents the output buffer from remaining uninitialized, thus denying a
+potential attack vector from an untrusted source.
+However, in the rare case where this scenario would be the outcome of a
+transmission or storage error, the decoder relies on the checksum to detect
+the error.
+
+In newer versions, this case is always detected and reported as a corruption error.
+
+
+Non-zeroes reserved bits
+------------------------
+
+**Last affected version**: v1.5.5
+
+**Produced by the reference compressor**: No
+
+The Sequences section of each block has a header, and one of its elements is a
+byte, which describes the compression mode of each symbol.
+This byte contains 2 reserved bits which must be set to zero.
+
+The reference decompressor up to v1.5.5 just ignores these 2 bits.
+This behavior has no consequence for the rest of the frame decoding process.
+
+In newer versions, the 2 reserved bits are actively checked for value zero,
+and the decoder reports a corruption error if they are not.

doc/educational_decoder/zstd_decompress.c

@@ -997,7 +997,8 @@ static void decompress_sequences(frame_context_t *const ctx,
                                  const size_t num_sequences);
 static sequence_command_t decode_sequence(sequence_states_t *const state,
                                           const u8 *const src,
-                                          i64 *const offset);
+                                          i64 *const offset,
+                                          int lastSequence);
 static void decode_seq_table(FSE_dtable *const table, istream_t *const in,
                                const seq_part_t type, const seq_mode_t mode);
 
@@ -1017,12 +1018,7 @@ static size_t decode_sequences(frame_context_t *const ctx, istream_t *in,
     // This is a variable size field using between 1 and 3 bytes. Let's call its
     // first byte byte0."
     u8 header = IO_read_bits(in, 8);
-    if (header == 0) {
-        // "There are no sequences. The sequence section stops there.
-        // Regenerated content is defined entirely by literals section."
-        *sequences = NULL;
-        return 0;
-    } else if (header < 128) {
+    if (header < 128) {
         // "Number_of_Sequences = byte0 . Uses 1 byte."
         num_sequences = header;
     } else if (header < 255) {
@@ -1033,6 +1029,12 @@ static size_t decode_sequences(frame_context_t *const ctx, istream_t *in,
         num_sequences = IO_read_bits(in, 16) + 0x7F00;
     }
 
+    if (num_sequences == 0) {
+        // "There are no sequences. The sequence section stops there."
+        *sequences = NULL;
+        return 0;
+    }
+
     *sequences = malloc(num_sequences * sizeof(sequence_command_t));
     if (!*sequences) {
         BAD_ALLOC();
@@ -1114,7 +1116,7 @@ static void decompress_sequences(frame_context_t *const ctx, istream_t *in,
 
     for (size_t i = 0; i < num_sequences; i++) {
         // Decode sequences one by one
-        sequences[i] = decode_sequence(&states, src, &bit_offset);
+        sequences[i] = decode_sequence(&states, src, &bit_offset, i==num_sequences-1);
     }
 
     if (bit_offset != 0) {
@@ -1125,7 +1127,8 @@ static void decompress_sequences(frame_context_t *const ctx, istream_t *in,
 // Decode a single sequence and update the state
 static sequence_command_t decode_sequence(sequence_states_t *const states,
                                           const u8 *const src,
-                                          i64 *const offset) {
+                                          i64 *const offset,
+                                          int lastSequence) {
     // "Each symbol is a code in its own context, which specifies Baseline and
     // Number_of_Bits to add. Codes are FSE compressed, and interleaved with raw
     // additional bits in the same bitstream."
@@ -1160,7 +1163,7 @@ static sequence_command_t decode_sequence(sequence_states_t *const states,
     // Literals_Length_State is updated, followed by Match_Length_State, and
     // then Offset_State."
     // If the stream is complete don't read bits to update state
-    if (*offset != 0) {
+    if (!lastSequence) {
         FSE_update_state(&states->ll_table, &states->ll_state, src, offset);
         FSE_update_state(&states->ml_table, &states->ml_state, src, offset);
         FSE_update_state(&states->of_table, &states->of_state, src, offset);
@@ -1210,7 +1213,7 @@ static void decode_seq_table(FSE_dtable *const table, istream_t *const in,
         break;
     }
     case seq_repeat:
-        // "Repeat_Mode : re-use distribution table from previous compressed
+        // "Repeat_Mode : reuse distribution table from previous compressed
         // block."
         // Nothing to do here, table will be unchanged
         if (!table->symbols) {
@@ -1399,7 +1402,7 @@ size_t ZSTD_get_decompressed_size(const void *src, const size_t src_len) {
 /******* END OUTPUT SIZE COUNTING *********************************************/
 
 /******* DICTIONARY PARSING ***************************************************/
-dictionary_t* create_dictionary() {
+dictionary_t* create_dictionary(void) {
     dictionary_t* const dict = calloc(1, sizeof(dictionary_t));
     if (!dict) {
         BAD_ALLOC();

doc/zstd_compression_format.md

@@ -16,7 +16,7 @@ Distribution of this document is unlimited.
 
 ### Version
 
-0.3.7 (2020-12-09)
+0.4.0 (2023-06-05)
 
 
 Introduction
@@ -470,6 +470,7 @@ This field uses 2 lowest bits of first byte, describing 4 different block types
         repeated `Regenerated_Size` times.
 - `Compressed_Literals_Block` - This is a standard Huffman-compressed block,
         starting with a Huffman tree description.
+        In this mode, there are at least 2 different literals represented in the Huffman tree description.
         See details below.
 - `Treeless_Literals_Block` - This is a Huffman-compressed block,
         using Huffman tree _from previous Huffman-compressed literals block_.
@@ -533,15 +534,20 @@ __`Size_Format` for `Compressed_Literals_Block` and `Treeless_Literals_Block`__
 Both `Compressed_Size` and `Regenerated_Size` fields follow __little-endian__ convention.
 Note: `Compressed_Size` __includes__ the size of the Huffman Tree description
 _when_ it is present.
+Note 2: `Compressed_Size` can never be `==0`.
+Even in single-stream scenario, assuming an empty content, it must be `>=1`,
+since it contains at least the final end bit flag.
+In 4-streams scenario, a valid `Compressed_Size` is necessarily `>= 10`
+(6 bytes for the jump table, + 4x1 bytes for the 4 streams).
 
-4 streams is superior to 1 stream in decompression speed,
+4 streams is faster than 1 stream in decompression speed,
 by exploiting instruction level parallelism.
 But it's also more expensive,
 costing on average ~7.3 bytes more than the 1 stream mode, mostly from the jump table.
 
 In general, use the 4 streams mode when there are more literals to decode,
 to favor higher decompression speeds.
-Beyond 1KB, the 4 streams mode is compulsory anyway.
+Note that beyond >1KB of literals, the 4 streams mode is compulsory.
 
 Note that a minimum of 6 bytes is required for the 4 streams mode.
 That's a technical minimum, but it's not recommended to employ the 4 streams mode
@@ -566,6 +572,7 @@ or from a dictionary.
 
 ### `Huffman_Tree_Description`
 This section is only present when `Literals_Block_Type` type is `Compressed_Literals_Block` (`2`).
+The tree describes the weights of all literals symbols that can be present in the literals block, at least 2 and up to 256.
 The format of the Huffman tree description can be found at [Huffman Tree description](#huffman-tree-description).
 The size of `Huffman_Tree_Description` is determined during decoding process,
 it must be used to determine where streams begin.
@@ -575,10 +582,10 @@ it must be used to determine where streams begin.
 ### Jump Table
 The Jump Table is only present when there are 4 Huffman-coded streams.
 
-Reminder : Huffman compressed data consists of either 1 or 4 Huffman-coded streams.
+Reminder : Huffman compressed data consists of either 1 or 4 streams.
 
 If only one stream is present, it is a single bitstream occupying the entire
-remaining portion of the literals block, encoded as described within
+remaining portion of the literals block, encoded as described in
 [Huffman-Coded Streams](#huffman-coded-streams).
 
 If there are four streams, `Literals_Section_Header` only provided
@@ -589,17 +596,18 @@ except for the last stream which may be up to 3 bytes smaller,
 to reach a total decompressed size as specified in `Regenerated_Size`.
 
 The compressed size of each stream is provided explicitly in the Jump Table.
-Jump Table is 6 bytes long, and consist of three 2-byte __little-endian__ fields,
+Jump Table is 6 bytes long, and consists of three 2-byte __little-endian__ fields,
 describing the compressed sizes of the first three streams.
-`Stream4_Size` is computed from total `Total_Streams_Size` minus sizes of other streams.
+`Stream4_Size` is computed from `Total_Streams_Size` minus sizes of other streams:
 
 `Stream4_Size = Total_Streams_Size - 6 - Stream1_Size - Stream2_Size - Stream3_Size`.
 
-Note: if `Stream1_Size + Stream2_Size + Stream3_Size > Total_Streams_Size`,
+`Stream4_Size` is necessarily `>= 1`. Therefore,
+if `Total_Streams_Size < Stream1_Size + Stream2_Size + Stream3_Size + 6 + 1`,
 data is considered corrupted.
 
 Each of these 4 bitstreams is then decoded independently as a Huffman-Coded stream,
-as described at [Huffman-Coded Streams](#huffman-coded-streams)
+as described in [Huffman-Coded Streams](#huffman-coded-streams)
 
 
 Sequences Section
@@ -642,13 +650,15 @@ __`Number_of_Sequences`__
 
 This is a variable size field using between 1 and 3 bytes.
 Let's call its first byte `byte0`.
-- `if (byte0 == 0)` : there are no sequences.
-            The sequence section stops there.
-            Decompressed content is defined entirely as Literals Section content.
-            The FSE tables used in `Repeat_Mode` aren't updated.
 - `if (byte0 < 128)` : `Number_of_Sequences = byte0` . Uses 1 byte.
-- `if (byte0 < 255)` : `Number_of_Sequences = ((byte0-128) << 8) + byte1` . Uses 2 bytes.
-- `if (byte0 == 255)`: `Number_of_Sequences = byte1 + (byte2<<8) + 0x7F00` . Uses 3 bytes.
+- `if (byte0 < 255)` : `Number_of_Sequences = ((byte0 - 0x80) << 8) + byte1`. Uses 2 bytes.
+            Note that the 2 bytes format fully overlaps the 1 byte format.
+- `if (byte0 == 255)`: `Number_of_Sequences = byte1 + (byte2<<8) + 0x7F00`. Uses 3 bytes.
+
+`if (Number_of_Sequences == 0)` : there are no sequences.
+            The sequence section stops immediately,
+            FSE tables used in `Repeat_Mode` aren't updated.
+            Block's decompressed content is defined solely by the Literals Section content.
 
 __Symbol compression modes__
 
@@ -919,7 +929,10 @@ There is an exception though, when current sequence's `literals_length = 0`.
 In this case, repeated offsets are shifted by one,
 so an `offset_value` of 1 means `Repeated_Offset2`,
 an `offset_value` of 2 means `Repeated_Offset3`,
-and an `offset_value` of 3 means `Repeated_Offset1 - 1_byte`.
+and an `offset_value` of 3 means `Repeated_Offset1 - 1`.
+
+In the final case, if `Repeated_Offset1 - 1` evaluates to 0, then the
+data is considered corrupted.
 
 For the first block, the starting offset history is populated with following values :
 `Repeated_Offset1`=1, `Repeated_Offset2`=4, `Repeated_Offset3`=8,
@@ -1073,7 +1086,8 @@ It depends on :
   Presuming an `Accuracy_Log` of 8,
   and presuming 100 probabilities points have already been distributed,
   the decoder may read any value from `0` to `256 - 100 + 1 == 157` (inclusive).
-  Therefore, it must read `log2sup(157) == 8` bits.
+  Therefore, it may read up to `log2sup(157) == 8` bits, where `log2sup(N)`
+  is the smallest integer `T` that satisfies `(1 << T) > N`.
 
 - Value decoded : small values use 1 less bit :
   __example__ :
@@ -1113,6 +1127,9 @@ When last symbol reaches cumulated total of `1 << Accuracy_Log`,
 decoding is complete.
 If the last symbol makes cumulated total go above `1 << Accuracy_Log`,
 distribution is considered corrupted.
+If this process results in a non-zero probability for a value outside of the
+valid range of values that the FSE table is defined for, even if that value is
+not used, then the data is considered corrupted.
 
 Then the decoder can tell how many bytes were used in this process,
 and how many symbols are present.
@@ -1176,9 +1193,9 @@ Baseline is assigned starting from the higher states using fewer bits,
 increasing at each state, then resuming at the first state,
 each state takes its allocated width from Baseline.
 
-| state value      |   1   |  39   |   77   |  84  |  122   |
 | state order      |   0   |   1   |    2   |   3  |    4   |
 | ---------------- | ----- | ----- | ------ | ---- | ------ |
+| state value      |   1   |  39   |   77   |  84  |  122   |
 | width            |  32   |  32   |   32   |  16  |   16   |
 | `Number_of_Bits` |   5   |   5   |    5   |   4  |    4   |
 | range number     |   2   |   4   |    6   |   0  |    1   |
@@ -1197,7 +1214,7 @@ Huffman Coding
 --------------
 Zstandard Huffman-coded streams are read backwards,
 similar to the FSE bitstreams.
-Therefore, to find the start of the bitstream, it is therefore to
+Therefore, to find the start of the bitstream, it is required to
 know the offset of the last byte of the Huffman-coded stream.
 
 After writing the last bit containing information, the compressor
@@ -1239,9 +1256,17 @@ Transformation from `Weight` to `Number_of_Bits` follows this formula :
 ```
 Number_of_Bits = Weight ? (Max_Number_of_Bits + 1 - Weight) : 0
 ```
-The last symbol's `Weight` is deduced from previously decoded ones,
-by completing to the nearest power of 2.
-This power of 2 gives `Max_Number_of_Bits`, the depth of the current tree.
+When a literal value is not present, it receives a `Weight` of 0.
+The least frequent symbol receives a `Weight` of 1.
+If no literal has a `Weight` of 1, then the data is considered corrupted.
+If there are not at least two literals with non-zero `Weight`, then the data
+is considered corrupted.
+The most frequent symbol receives a `Weight` anywhere between 1 and 11 (max).
+The last symbol's `Weight` is deduced from previously retrieved Weights,
+by completing to the nearest power of 2. It's necessarily non 0.
+If it's not possible to reach a clean power of 2 with a single `Weight` value,
+the Huffman Tree Description is considered invalid.
+This final power of 2 gives `Max_Number_of_Bits`, the depth of the current tree.
 `Max_Number_of_Bits` must be <= 11,
 otherwise the representation is considered corrupted.
 
@@ -1254,7 +1279,7 @@ Let's presume the following Huffman tree must be described :
 
 The tree depth is 4, since its longest elements uses 4 bits
 (longest elements are the one with smallest frequency).
-Value `5` will not be listed, as it can be determined from values for 0-4,
+Literal value `5` will not be listed, as it can be determined from previous values 0-4,
 nor will values above `5` as they are all 0.
 Values from `0` to `4` will be listed using `Weight` instead of `Number_of_Bits`.
 Weight formula is :
@@ -1274,7 +1299,7 @@ The `Weight` of `5` can be determined by advancing to the next power of 2.
 The sum of `2^(Weight-1)` (excluding 0's) is :
 `8 + 4 + 2 + 0 + 1 = 15`.
 Nearest larger power of 2 value is 16.
-Therefore, `Max_Number_of_Bits = 4` and `Weight[5] = 16-15 = 1`.
+Therefore, `Max_Number_of_Bits = 4` and `Weight[5] = log_2(16 - 15) + 1 = 1`.
 
 #### Huffman Tree header
 
@@ -1336,6 +1361,9 @@ If updating state after decoding a symbol would require more bits than
 remain in the stream, it is assumed that extra bits are 0.  Then,
 symbols for each of the final states are decoded and the process is complete.
 
+If this process would produce more weights than the maximum number of decoded
+weights (255), then the data is considered corrupted.
+
 #### Conversion from weights to Huffman prefix codes
 
 All present symbols shall now have a `Weight` value.
@@ -1683,6 +1711,9 @@ or at least provide a meaningful error code explaining for which reason it canno
 
 Version changes
 ---------------
+- 0.4.0 : fixed imprecise behavior for nbSeq==0, detected by Igor Pavlov
+- 0.3.9 : clarifications for Huffman-compressed literal sizes.
+- 0.3.8 : clarifications for Huffman Blocks and Huffman Tree descriptions.
 - 0.3.7 : clarifications for Repeat_Offsets, matching RFC8878
 - 0.3.6 : clarifications for Dictionary_ID
 - 0.3.5 : clarifications for Block_Maximum_Size

doc/zstd_manual.html

@@ -1,10 +1,10 @@
 <html>
 <head>
 <meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
-<title>zstd 1.5.4 Manual</title>
+<title>zstd 1.5.6 Manual</title>
 </head>
 <body>
-<h1>zstd 1.5.4 Manual</h1>
+<h1>zstd 1.5.6 Manual</h1>
 <hr>
 <a name="Contents"></a><h2>Contents</h2>
 <ol>
@@ -22,15 +22,15 @@
 <li><a href="#Chapter12">Dictionary helper functions</a></li>
 <li><a href="#Chapter13">Advanced dictionary and prefix API (Requires v1.4.0+)</a></li>
 <li><a href="#Chapter14">experimental API (static linking only)</a></li>
-<li><a href="#Chapter15">Frame size functions</a></li>
+<li><a href="#Chapter15">Frame header and size functions</a></li>
 <li><a href="#Chapter16">Memory management</a></li>
 <li><a href="#Chapter17">Advanced compression functions</a></li>
 <li><a href="#Chapter18">Advanced decompression functions</a></li>
 <li><a href="#Chapter19">Advanced streaming functions</a></li>
-<li><a href="#Chapter20">Buffer-less and synchronous inner streaming functions</a></li>
+<li><a href="#Chapter20">Buffer-less and synchronous inner streaming functions (DEPRECATED)</a></li>
 <li><a href="#Chapter21">Buffer-less streaming compression (synchronous mode)</a></li>
 <li><a href="#Chapter22">Buffer-less streaming decompression (synchronous mode)</a></li>
-<li><a href="#Chapter23">Block level API</a></li>
+<li><a href="#Chapter23">Block level API (DEPRECATED)</a></li>
 </ol>
 <hr>
 <a name="Chapter1"></a><h2>Introduction</h2><pre>
@@ -80,7 +80,8 @@
                 const void* src, size_t srcSize,
                       int compressionLevel);
 </b><p>  Compresses `src` content as a single zstd compressed frame into already allocated `dst`.
-  Hint : compression runs faster if `dstCapacity` >=  `ZSTD_compressBound(srcSize)`.
+  NOTE: Providing `dstCapacity >= ZSTD_compressBound(srcSize)` guarantees that zstd will have
+        enough space to successfully compress the data.
   @return : compressed size written into `dst` (<= `dstCapacity),
             or an error code if it fails (which can be tested using ZSTD_isError()). 
 </p></pre><BR>
@@ -155,7 +156,7 @@ unsigned long long ZSTD_getDecompressedSize(const void* src, size_t srcSize);
  * for example to size a static array on stack.
  * Will produce constant value 0 if srcSize too large.
  */
-#define ZSTD_MAX_INPUT_SIZE ((sizeof(size_t)==8) ? 0xFF00FF00FF00FF00LLU : 0xFF00FF00U)
+#define ZSTD_MAX_INPUT_SIZE ((sizeof(size_t)==8) ? 0xFF00FF00FF00FF00ULL : 0xFF00FF00U)
 #define ZSTD_COMPRESSBOUND(srcSize)   (((size_t)(srcSize) >= ZSTD_MAX_INPUT_SIZE) ? 0 : (srcSize) + ((srcSize)>>8) + (((srcSize) < (128<<10)) ? (((128<<10) - (srcSize)) >> 11) </b>/* margin, from 64 to 0 */ : 0))  /* this formula ensures that bound(A) + bound(B) <= bound(A+B) as long as A and B >= 128 KB */<b>
 size_t ZSTD_compressBound(size_t srcSize); </b>/*!< maximum compressed size in worst case single-pass scenario */<b>
 </b>/* ZSTD_isError() :<b>
@@ -173,7 +174,7 @@ int         ZSTD_defaultCLevel(void);           </b>/*!< default compression lev
 
 <h3>Compression context</h3><pre>  When compressing many times,
   it is recommended to allocate a context just once,
-  and re-use it for each successive compression operation.
+  and reuse it for each successive compression operation.
   This will make workload friendlier for system's memory.
   Note : re-using context is just a speed / resource optimization.
          It doesn't change the compression ratio, which remains identical.
@@ -189,9 +190,9 @@ size_t     ZSTD_freeCCtx(ZSTD_CCtx* cctx);  </b>/* accept NULL pointer */<b>
                    const void* src, size_t srcSize,
                          int compressionLevel);
 </b><p>  Same as ZSTD_compress(), using an explicit ZSTD_CCtx.
-  Important : in order to behave similarly to `ZSTD_compress()`,
-  this function compresses at requested compression level,
-  __ignoring any other parameter__ .
+  Important : in order to mirror `ZSTD_compress()` behavior,
+  this function compresses at the requested compression level,
+  __ignoring any other advanced parameter__ .
   If any advanced parameter was set using the advanced API,
   they will all be reset. Only `compressionLevel` remains.
  
@@ -199,7 +200,7 @@ size_t     ZSTD_freeCCtx(ZSTD_CCtx* cctx);  </b>/* accept NULL pointer */<b>
 
 <h3>Decompression context</h3><pre>  When decompressing many times,
   it is recommended to allocate a context only once,
-  and re-use it for each successive compression operation.
+  and reuse it for each successive compression operation.
   This will make workload friendlier for system's memory.
   Use one context per thread for parallel execution. 
 </pre><b><pre>typedef struct ZSTD_DCtx_s ZSTD_DCtx;
@@ -211,7 +212,7 @@ size_t     ZSTD_freeDCtx(ZSTD_DCtx* dctx);  </b>/* accept NULL pointer */<b>
                      const void* src, size_t srcSize);
 </b><p>  Same as ZSTD_decompress(),
   requires an allocated ZSTD_DCtx.
-  Compatible with sticky parameters.
+  Compatible with sticky parameters (see below).
  
 </p></pre><BR>
 
@@ -295,6 +296,19 @@ size_t     ZSTD_freeDCtx(ZSTD_DCtx* dctx);  </b>/* accept NULL pointer */<b>
                               * The higher the value of selected strategy, the more complex it is,
                               * resulting in stronger and slower compression.
                               * Special: value 0 means "use default strategy". */
+
+    ZSTD_c_targetCBlockSize=130, </b>/* v1.5.6+<b>
+                                  * Attempts to fit compressed block size into approximatively targetCBlockSize.
+                                  * Bound by ZSTD_TARGETCBLOCKSIZE_MIN and ZSTD_TARGETCBLOCKSIZE_MAX.
+                                  * Note that it's not a guarantee, just a convergence target (default:0).
+                                  * No target when targetCBlockSize == 0.
+                                  * This is helpful in low bandwidth streaming environments to improve end-to-end latency,
+                                  * when a client can make use of partial documents (a prominent example being Chrome).
+                                  * Note: this parameter is stable since v1.5.6.
+                                  * It was present as an experimental parameter in earlier versions,
+                                  * but it's not recommended using it with earlier library versions
+                                  * due to massive performance regressions.
+                                  */
     </b>/* LDM mode parameters */<b>
     ZSTD_c_enableLongDistanceMatching=160, </b>/* Enable long distance matching.<b>
                                      * This parameter is designed to improve compression ratio
@@ -374,7 +388,6 @@ size_t     ZSTD_freeDCtx(ZSTD_DCtx* dctx);  </b>/* accept NULL pointer */<b>
      * ZSTD_c_forceMaxWindow
      * ZSTD_c_forceAttachDict
      * ZSTD_c_literalCompressionMode
-     * ZSTD_c_targetCBlockSize
      * ZSTD_c_srcSizeHint
      * ZSTD_c_enableDedicatedDictSearch
      * ZSTD_c_stableInBuffer
@@ -395,7 +408,7 @@ size_t     ZSTD_freeDCtx(ZSTD_DCtx* dctx);  </b>/* accept NULL pointer */<b>
      ZSTD_c_experimentalParam3=1000,
      ZSTD_c_experimentalParam4=1001,
      ZSTD_c_experimentalParam5=1002,
-     ZSTD_c_experimentalParam6=1003,
+     </b>/* was ZSTD_c_experimentalParam6=1003; is now ZSTD_c_targetCBlockSize */<b>
      ZSTD_c_experimentalParam7=1004,
      ZSTD_c_experimentalParam8=1005,
      ZSTD_c_experimentalParam9=1006,
@@ -482,11 +495,13 @@ size_t     ZSTD_freeDCtx(ZSTD_DCtx* dctx);  </b>/* accept NULL pointer */<b>
                        void* dst, size_t dstCapacity,
                  const void* src, size_t srcSize);
 </b><p>  Behave the same as ZSTD_compressCCtx(), but compression parameters are set using the advanced API.
+  (note that this entry point doesn't even expose a compression level parameter).
   ZSTD_compress2() always starts a new frame.
   Should cctx hold data from a previously unfinished frame, everything about it is forgotten.
   - Compression parameters are pushed into CCtx before starting compression, using ZSTD_CCtx_set*()
   - The function is always blocking, returns when compression is completed.
-  Hint : compression runs faster if `dstCapacity` >=  `ZSTD_compressBound(srcSize)`.
+  NOTE: Providing `dstCapacity >= ZSTD_compressBound(srcSize)` guarantees that zstd will have
+        enough space to successfully compress the data, though it is possible it fails for other reasons.
  @return : compressed size written into `dst` (<= `dstCapacity),
            or an error code if it fails (which can be tested using ZSTD_isError()).
  
@@ -511,6 +526,7 @@ size_t     ZSTD_freeDCtx(ZSTD_DCtx* dctx);  </b>/* accept NULL pointer */<b>
      * ZSTD_d_forceIgnoreChecksum
      * ZSTD_d_refMultipleDDicts
      * ZSTD_d_disableHuffmanAssembly
+     * ZSTD_d_maxBlockSize
      * Because they are not stable, it's necessary to define ZSTD_STATIC_LINKING_ONLY to access them.
      * note : never ever use experimentalParam? names directly
      */
@@ -518,7 +534,8 @@ size_t     ZSTD_freeDCtx(ZSTD_DCtx* dctx);  </b>/* accept NULL pointer */<b>
      ZSTD_d_experimentalParam2=1001,
      ZSTD_d_experimentalParam3=1002,
      ZSTD_d_experimentalParam4=1003,
-     ZSTD_d_experimentalParam5=1004
+     ZSTD_d_experimentalParam5=1004,
+     ZSTD_d_experimentalParam6=1005
 
 } ZSTD_dParameter;
 </b></pre><BR>
@@ -566,14 +583,14 @@ size_t     ZSTD_freeDCtx(ZSTD_DCtx* dctx);  </b>/* accept NULL pointer */<b>
   A ZSTD_CStream object is required to track streaming operation.
   Use ZSTD_createCStream() and ZSTD_freeCStream() to create/release resources.
   ZSTD_CStream objects can be reused multiple times on consecutive compression operations.
-  It is recommended to re-use ZSTD_CStream since it will play nicer with system's memory, by re-using already allocated memory.
+  It is recommended to reuse ZSTD_CStream since it will play nicer with system's memory, by re-using already allocated memory.
 
   For parallel execution, use one separate ZSTD_CStream per thread.
 
   note : since v1.3.0, ZSTD_CStream and ZSTD_CCtx are the same thing.
 
   Parameters are sticky : when starting a new compression on the same context,
-  it will re-use the same sticky parameters as previous compression session.
+  it will reuse the same sticky parameters as previous compression session.
   When in doubt, it's recommended to fully initialize the context before usage.
   Use ZSTD_CCtx_reset() to reset the context and ZSTD_CCtx_setParameter(),
   ZSTD_CCtx_setPledgedSrcSize(), or ZSTD_CCtx_loadDictionary() and friends to
@@ -664,6 +681,11 @@ size_t ZSTD_freeCStream(ZSTD_CStream* zcs);  </b>/* accept NULL pointer */<b>
             only ZSTD_e_end or ZSTD_e_flush operations are allowed.
             Before starting a new compression job, or changing compression parameters,
             it is required to fully flush internal buffers.
+  - note: if an operation ends with an error, it may leave @cctx in an undefined state.
+          Therefore, it's UB to invoke ZSTD_compressStream2() of ZSTD_compressStream() on such a state.
+          In order to be re-employed after an error, a state must be reset,
+          which can be done explicitly (ZSTD_CCtx_reset()),
+          or is sometimes implied by methods starting a new compression job (ZSTD_initCStream(), ZSTD_compressCCtx())
  
 </p></pre><BR>
 
@@ -696,7 +718,7 @@ size_t ZSTD_endStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output);
 <a name="Chapter9"></a><h2>Streaming decompression - HowTo</h2><pre>
   A ZSTD_DStream object is required to track streaming operations.
   Use ZSTD_createDStream() and ZSTD_freeDStream() to create/release resources.
-  ZSTD_DStream objects can be re-used multiple times.
+  ZSTD_DStream objects can be reused multiple times.
 
   Use ZSTD_initDStream() to start a new decompression operation.
  @return : recommended first input size
@@ -749,6 +771,12 @@ size_t ZSTD_freeDStream(ZSTD_DStream* zds);  </b>/* accept NULL pointer */<b>
  @return : 0 when a frame is completely decoded and fully flushed,
            or an error code, which can be tested using ZSTD_isError(),
            or any other value > 0, which means there is some decoding or flushing to do to complete current frame.
+
+ Note: when an operation returns with an error code, the @zds state may be left in undefined state.
+       It's UB to invoke `ZSTD_decompressStream()` on such a state.
+       In order to re-use such a state, it must be first reset,
+       which can be done explicitly (`ZSTD_DCtx_reset()`),
+       or is implied for operations starting some new decompression job (`ZSTD_initDStream`, `ZSTD_decompressDCtx()`, `ZSTD_decompress_usingDict()`)
  
 </p></pre><BR>
 
@@ -866,9 +894,11 @@ size_t ZSTD_freeDStream(ZSTD_DStream* zds);  </b>/* accept NULL pointer */<b>
 
 <a name="Chapter13"></a><h2>Advanced dictionary and prefix API (Requires v1.4.0+)</h2><pre>
  This API allows dictionaries to be used with ZSTD_compress2(),
- ZSTD_compressStream2(), and ZSTD_decompressDCtx(). Dictionaries are sticky, and
- only reset with the context is reset with ZSTD_reset_parameters or
- ZSTD_reset_session_and_parameters. Prefixes are single-use.
+ ZSTD_compressStream2(), and ZSTD_decompressDCtx().
+ Dictionaries are sticky, they remain valid when same context is reused,
+ they only reset when the context is reset
+ with ZSTD_reset_parameters or ZSTD_reset_session_and_parameters.
+ In contrast, Prefixes are single-use.
 <BR></pre>
 
 <pre><b>size_t ZSTD_CCtx_loadDictionary(ZSTD_CCtx* cctx, const void* dict, size_t dictSize);
@@ -888,7 +918,11 @@ size_t ZSTD_freeDStream(ZSTD_DStream* zds);  </b>/* accept NULL pointer */<b>
            Use experimental ZSTD_CCtx_loadDictionary_byReference() to reference content instead.
            In such a case, dictionary buffer must outlive its users.
   Note 4 : Use ZSTD_CCtx_loadDictionary_advanced()
-           to precisely select how dictionary content must be interpreted. 
+           to precisely select how dictionary content must be interpreted.
+  Note 5 : This method does not benefit from LDM (long distance mode).
+           If you want to employ LDM on some large dictionary content,
+           prefer employing ZSTD_CCtx_refPrefix() described below.
+ 
 </p></pre><BR>
 
 <pre><b>size_t ZSTD_CCtx_refCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict);
@@ -912,6 +946,7 @@ size_t ZSTD_freeDStream(ZSTD_DStream* zds);  </b>/* accept NULL pointer */<b>
   Decompression will need same prefix to properly regenerate data.
   Compressing with a prefix is similar in outcome as performing a diff and compressing it,
   but performs much faster, especially during decompression (compression speed is tunable with compression level).
+  This method is compatible with LDM (long distance mode).
  @result : 0, or an error code (which can be tested with ZSTD_isError()).
   Special: Adding any prefix (including NULL) invalidates any previous prefix or dictionary
   Note 1 : Prefix buffer is referenced. It **must** outlive compression.
@@ -1146,7 +1181,7 @@ size_t ZSTD_sizeof_DDict(const ZSTD_DDict* ddict);
   ZSTD_ps_disable = 2       </b>/* Do not use the feature */<b>
 } ZSTD_paramSwitch_e;
 </b></pre><BR>
-<a name="Chapter15"></a><h2>Frame size functions</h2><pre></pre>
+<a name="Chapter15"></a><h2>Frame header and size functions</h2><pre></pre>
 
 <pre><b>ZSTDLIB_STATIC_API unsigned long long ZSTD_findDecompressedSize(const void* src, size_t srcSize);
 </b><p>  `src` should point to the start of a series of ZSTD encoded and/or skippable frames
@@ -1192,6 +1227,31 @@ size_t ZSTD_sizeof_DDict(const ZSTD_DDict* ddict);
            or an error code (if srcSize is too small) 
 </p></pre><BR>
 
+<pre><b>typedef enum { ZSTD_frame, ZSTD_skippableFrame } ZSTD_frameType_e;
+</b></pre><BR>
+<pre><b>typedef struct {
+    unsigned long long frameContentSize; </b>/* if == ZSTD_CONTENTSIZE_UNKNOWN, it means this field is not available. 0 means "empty" */<b>
+    unsigned long long windowSize;       </b>/* can be very large, up to <= frameContentSize */<b>
+    unsigned blockSizeMax;
+    ZSTD_frameType_e frameType;          </b>/* if == ZSTD_skippableFrame, frameContentSize is the size of skippable content */<b>
+    unsigned headerSize;
+    unsigned dictID;
+    unsigned checksumFlag;
+    unsigned _reserved1;
+    unsigned _reserved2;
+} ZSTD_frameHeader;
+</b></pre><BR>
+<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_getFrameHeader(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize);   </b>/**< doesn't consume input */<b>
+</b>/*! ZSTD_getFrameHeader_advanced() :<b>
+ *  same as ZSTD_getFrameHeader(),
+ *  with added capability to select a format (like ZSTD_f_zstd1_magicless) */
+ZSTDLIB_STATIC_API size_t ZSTD_getFrameHeader_advanced(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize, ZSTD_format_e format);
+</b><p>  decode Frame Header, or requires larger `srcSize`.
+ @return : 0, `zfhPtr` is correctly filled,
+          >0, `srcSize` is too small, value is wanted `srcSize` amount,
+           or an error code, which can be tested using ZSTD_isError() 
+</p></pre><BR>
+
 <pre><b>ZSTDLIB_STATIC_API size_t ZSTD_decompressionMargin(const void* src, size_t srcSize);
 </b><p> Zstd supports in-place decompression, where the input and output buffers overlap.
  In this case, the output buffer must be at least (Margin + Output_Size) bytes large,
@@ -1352,58 +1412,61 @@ ZSTD_compressSequences( ZSTD_CCtx* cctx, void* dst, size_t dstSize,
 
 <a name="Chapter16"></a><h2>Memory management</h2><pre></pre>
 
-<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_estimateCCtxSize(int compressionLevel);
+<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_estimateCCtxSize(int maxCompressionLevel);
 ZSTDLIB_STATIC_API size_t ZSTD_estimateCCtxSize_usingCParams(ZSTD_compressionParameters cParams);
 ZSTDLIB_STATIC_API size_t ZSTD_estimateCCtxSize_usingCCtxParams(const ZSTD_CCtx_params* params);
 ZSTDLIB_STATIC_API size_t ZSTD_estimateDCtxSize(void);
 </b><p>  These functions make it possible to estimate memory usage
   of a future {D,C}Ctx, before its creation.
+  This is useful in combination with ZSTD_initStatic(),
+  which makes it possible to employ a static buffer for ZSTD_CCtx* state.
 
   ZSTD_estimateCCtxSize() will provide a memory budget large enough
-  for any compression level up to selected one.
-  Note : Unlike ZSTD_estimateCStreamSize*(), this estimate
-         does not include space for a window buffer.
-         Therefore, the estimation is only guaranteed for single-shot compressions, not streaming.
+  to compress data of any size using one-shot compression ZSTD_compressCCtx() or ZSTD_compress2()
+  associated with any compression level up to max specified one.
   The estimate will assume the input may be arbitrarily large,
   which is the worst case.
 
+  Note that the size estimation is specific for one-shot compression,
+  it is not valid for streaming (see ZSTD_estimateCStreamSize*())
+  nor other potential ways of using a ZSTD_CCtx* state.
+
   When srcSize can be bound by a known and rather "small" value,
-  this fact can be used to provide a tighter estimation
-  because the CCtx compression context will need less memory.
-  This tighter estimation can be provided by more advanced functions
+  this knowledge can be used to provide a tighter budget estimation
+  because the ZSTD_CCtx* state will need less memory for small inputs.
+  This tighter estimation can be provided by employing more advanced functions
   ZSTD_estimateCCtxSize_usingCParams(), which can be used in tandem with ZSTD_getCParams(),
   and ZSTD_estimateCCtxSize_usingCCtxParams(), which can be used in tandem with ZSTD_CCtxParams_setParameter().
   Both can be used to estimate memory using custom compression parameters and arbitrary srcSize limits.
 
   Note : only single-threaded compression is supported.
   ZSTD_estimateCCtxSize_usingCCtxParams() will return an error code if ZSTD_c_nbWorkers is >= 1.
-
-  Note 2 : ZSTD_estimateCCtxSize* functions are not compatible with the Block-Level Sequence Producer API at this time.
-  Size estimates assume that no external sequence producer is registered.
  
 </p></pre><BR>
 
-<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_estimateCStreamSize(int compressionLevel);
+<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_estimateCStreamSize(int maxCompressionLevel);
 ZSTDLIB_STATIC_API size_t ZSTD_estimateCStreamSize_usingCParams(ZSTD_compressionParameters cParams);
 ZSTDLIB_STATIC_API size_t ZSTD_estimateCStreamSize_usingCCtxParams(const ZSTD_CCtx_params* params);
-ZSTDLIB_STATIC_API size_t ZSTD_estimateDStreamSize(size_t windowSize);
+ZSTDLIB_STATIC_API size_t ZSTD_estimateDStreamSize(size_t maxWindowSize);
 ZSTDLIB_STATIC_API size_t ZSTD_estimateDStreamSize_fromFrame(const void* src, size_t srcSize);
-</b><p>  ZSTD_estimateCStreamSize() will provide a budget large enough for any compression level up to selected one.
-  It will also consider src size to be arbitrarily "large", which is worst case.
+</b><p>  ZSTD_estimateCStreamSize() will provide a memory budget large enough for streaming compression
+  using any compression level up to the max specified one.
+  It will also consider src size to be arbitrarily "large", which is a worst case scenario.
   If srcSize is known to always be small, ZSTD_estimateCStreamSize_usingCParams() can provide a tighter estimation.
   ZSTD_estimateCStreamSize_usingCParams() can be used in tandem with ZSTD_getCParams() to create cParams from compressionLevel.
   ZSTD_estimateCStreamSize_usingCCtxParams() can be used in tandem with ZSTD_CCtxParams_setParameter(). Only single-threaded compression is supported. This function will return an error code if ZSTD_c_nbWorkers is >= 1.
   Note : CStream size estimation is only correct for single-threaded compression.
-  ZSTD_DStream memory budget depends on window Size.
+  ZSTD_estimateCStreamSize_usingCCtxParams() will return an error code if ZSTD_c_nbWorkers is >= 1.
+  Note 2 : ZSTD_estimateCStreamSize* functions are not compatible with the Block-Level Sequence Producer API at this time.
+  Size estimates assume that no external sequence producer is registered.
+
+  ZSTD_DStream memory budget depends on frame's window Size.
   This information can be passed manually, using ZSTD_estimateDStreamSize,
   or deducted from a valid frame Header, using ZSTD_estimateDStreamSize_fromFrame();
+  Any frame requesting a window size larger than max specified one will be rejected.
   Note : if streaming is init with function ZSTD_init?Stream_usingDict(),
          an internal ?Dict will be created, which additional size is not estimated here.
          In this case, get total size by adding ZSTD_estimate?DictSize
-  Note 2 : only single-threaded compression is supported.
-  ZSTD_estimateCStreamSize_usingCCtxParams() will return an error code if ZSTD_c_nbWorkers is >= 1.
-  Note 3 : ZSTD_estimateCStreamSize* functions are not compatible with the Block-Level Sequence Producer API at this time.
-  Size estimates assume that no external sequence producer is registered.
  
 </p></pre><BR>
 
@@ -1505,10 +1568,24 @@ ZSTDLIB_STATIC_API size_t ZSTD_CCtx_refThreadPool(ZSTD_CCtx* cctx, ZSTD_threadPo
 </p></pre><BR>
 
 <pre><b>ZSTDLIB_STATIC_API size_t ZSTD_CCtx_setCParams(ZSTD_CCtx* cctx, ZSTD_compressionParameters cparams);
-</b><p>  Set all parameters provided within @cparams into the working @cctx.
+</b><p>  Set all parameters provided within @p cparams into the working @p cctx.
   Note : if modifying parameters during compression (MT mode only),
          note that changes to the .windowLog parameter will be ignored.
- @return 0 on success, or an error code (can be checked with ZSTD_isError()) 
+ @return 0 on success, or an error code (can be checked with ZSTD_isError()).
+         On failure, no parameters are updated.
+ 
+</p></pre><BR>
+
+<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_CCtx_setFParams(ZSTD_CCtx* cctx, ZSTD_frameParameters fparams);
+</b><p>  Set all parameters provided within @p fparams into the working @p cctx.
+ @return 0 on success, or an error code (can be checked with ZSTD_isError()).
+ 
+</p></pre><BR>
+
+<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_CCtx_setParams(ZSTD_CCtx* cctx, ZSTD_parameters params);
+</b><p>  Set all parameters provided within @p params into the working @p cctx.
+ @return 0 on success, or an error code (can be checked with ZSTD_isError()).
+ 
 </p></pre><BR>
 
 <pre><b>ZSTD_DEPRECATED("use ZSTD_compress2")
@@ -1754,12 +1831,9 @@ size_t ZSTD_initCStream_advanced(ZSTD_CStream* zcs,
         const void* dict, size_t dictSize,
               ZSTD_parameters params,
               unsigned long long pledgedSrcSize);
-</b><p> This function is DEPRECATED, and is approximately equivalent to:
+</b><p> This function is DEPRECATED, and is equivalent to:
      ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
-     // Pseudocode: Set each zstd parameter and leave the rest as-is.
-     for ((param, value) : params) {
-         ZSTD_CCtx_setParameter(zcs, param, value);
-     }
+     ZSTD_CCtx_setParams(zcs, params);
      ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize);
      ZSTD_CCtx_loadDictionary(zcs, dict, dictSize);
 
@@ -1788,12 +1862,9 @@ size_t ZSTD_initCStream_usingCDict_advanced(ZSTD_CStream* zcs,
                    const ZSTD_CDict* cdict,
                          ZSTD_frameParameters fParams,
                          unsigned long long pledgedSrcSize);
-</b><p>   This function is DEPRECATED, and is approximately equivalent to:
+</b><p>   This function is DEPRECATED, and is equivalent to:
      ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
-     // Pseudocode: Set each zstd frame parameter and leave the rest as-is.
-     for ((fParam, value) : fParams) {
-         ZSTD_CCtx_setParameter(zcs, fParam, value);
-     }
+     ZSTD_CCtx_setFParams(zcs, fParams);
      ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize);
      ZSTD_CCtx_refCDict(zcs, cdict);
 
@@ -1815,7 +1886,7 @@ size_t ZSTD_resetCStream(ZSTD_CStream* zcs, unsigned long long pledgedSrcSize);
        explicitly specified.
 
   start a new frame, using same parameters from previous frame.
-  This is typically useful to skip dictionary loading stage, since it will re-use it in-place.
+  This is typically useful to skip dictionary loading stage, since it will reuse it in-place.
   Note that zcs must be init at least once before using ZSTD_resetCStream().
   If pledgedSrcSize is not known at reset time, use macro ZSTD_CONTENTSIZE_UNKNOWN.
   If pledgedSrcSize > 0, its value must be correct, as it will be written in header, and controlled at the end.
@@ -1876,21 +1947,69 @@ ZSTDLIB_STATIC_API size_t ZSTD_resetDStream(ZSTD_DStream* zds);
 </b><p>
      ZSTD_DCtx_reset(zds, ZSTD_reset_session_only);
 
- re-use decompression parameters from previous init; saves dictionary loading
+ reuse decompression parameters from previous init; saves dictionary loading
  
 </p></pre><BR>
 
-<a name="Chapter20"></a><h2>Buffer-less and synchronous inner streaming functions</h2><pre>
-  This is an advanced API, giving full control over buffer management, for users which need direct control over memory.
-  But it's also a complex one, with several restrictions, documented below.
-  Prefer normal streaming API for an easier experience.
+<pre><b>ZSTDLIB_STATIC_API void
+ZSTD_registerSequenceProducer(
+  ZSTD_CCtx* cctx,
+  void* sequenceProducerState,
+  ZSTD_sequenceProducer_F sequenceProducer
+);
+</b><p> Instruct zstd to use a block-level external sequence producer function.
+
+ The sequenceProducerState must be initialized by the caller, and the caller is
+ responsible for managing its lifetime. This parameter is sticky across
+ compressions. It will remain set until the user explicitly resets compression
+ parameters.
+
+ Sequence producer registration is considered to be an "advanced parameter",
+ part of the "advanced API". This means it will only have an effect on compression
+ APIs which respect advanced parameters, such as compress2() and compressStream2().
+ Older compression APIs such as compressCCtx(), which predate the introduction of
+ "advanced parameters", will ignore any external sequence producer setting.
+
+ The sequence producer can be "cleared" by registering a NULL function pointer. This
+ removes all limitations described above in the "LIMITATIONS" section of the API docs.
+
+ The user is strongly encouraged to read the full API documentation (above) before
+ calling this function. 
+</p></pre><BR>
+
+<pre><b>ZSTDLIB_STATIC_API void
+ZSTD_CCtxParams_registerSequenceProducer(
+  ZSTD_CCtx_params* params,
+  void* sequenceProducerState,
+  ZSTD_sequenceProducer_F sequenceProducer
+);
+</b><p> Same as ZSTD_registerSequenceProducer(), but operates on ZSTD_CCtx_params.
+ This is used for accurate size estimation with ZSTD_estimateCCtxSize_usingCCtxParams(),
+ which is needed when creating a ZSTD_CCtx with ZSTD_initStaticCCtx().
+
+ If you are using the external sequence producer API in a scenario where ZSTD_initStaticCCtx()
+ is required, then this function is for you. Otherwise, you probably don't need it.
+
+ See tests/zstreamtest.c for example usage. 
+</p></pre><BR>
+
+<a name="Chapter20"></a><h2>Buffer-less and synchronous inner streaming functions (DEPRECATED)</h2><pre>
+  This API is deprecated, and will be removed in a future version.
+  It allows streaming (de)compression with user allocated buffers.
+  However, it is hard to use, and not as well tested as the rest of
+  our API.
+
+  Please use the normal streaming API instead: ZSTD_compressStream2,
+  and ZSTD_decompressStream.
+  If there is functionality that you need, but it doesn't provide,
+  please open an issue on our GitHub.
  
 <BR></pre>
 
 <a name="Chapter21"></a><h2>Buffer-less streaming compression (synchronous mode)</h2><pre>
   A ZSTD_CCtx object is required to track streaming operations.
   Use ZSTD_createCCtx() / ZSTD_freeCCtx() to manage resource.
-  ZSTD_CCtx object can be re-used multiple times within successive compression operations.
+  ZSTD_CCtx object can be reused multiple times within successive compression operations.
 
   Start by initializing a context.
   Use ZSTD_compressBegin(), or ZSTD_compressBegin_usingDict() for dictionary compression.
@@ -1911,11 +2030,14 @@ ZSTDLIB_STATIC_API size_t ZSTD_resetDStream(ZSTD_DStream* zds);
   It's possible to use srcSize==0, in which case, it will write a final empty block to end the frame.
   Without last block mark, frames are considered unfinished (hence corrupted) by compliant decoders.
 
-  `ZSTD_CCtx` object can be re-used (ZSTD_compressBegin()) to compress again.
+  `ZSTD_CCtx` object can be reused (ZSTD_compressBegin()) to compress again.
 <BR></pre>
 
-<h3>Buffer-less streaming compression functions</h3><pre></pre><b><pre>ZSTDLIB_STATIC_API size_t ZSTD_compressBegin(ZSTD_CCtx* cctx, int compressionLevel);
+<h3>Buffer-less streaming compression functions</h3><pre></pre><b><pre>ZSTD_DEPRECATED("The buffer-less API is deprecated in favor of the normal streaming API. See docs.")
+ZSTDLIB_STATIC_API size_t ZSTD_compressBegin(ZSTD_CCtx* cctx, int compressionLevel);
+ZSTD_DEPRECATED("The buffer-less API is deprecated in favor of the normal streaming API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_compressBegin_usingDict(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, int compressionLevel);
+ZSTD_DEPRECATED("The buffer-less API is deprecated in favor of the normal streaming API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_compressBegin_usingCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict); </b>/**< note: fails if cdict==NULL */<b>
 </pre></b><BR>
 <pre><b>size_t ZSTD_copyCCtx(ZSTD_CCtx* cctx, const ZSTD_CCtx* preparedCCtx, unsigned long long pledgedSrcSize); </b>/**<  note: if pledgedSrcSize is not known, use ZSTD_CONTENTSIZE_UNKNOWN */<b>
@@ -1925,7 +2047,7 @@ ZSTDLIB_STATIC_API size_t ZSTD_compressBegin_usingCDict(ZSTD_CCtx* cctx, const Z
 <a name="Chapter22"></a><h2>Buffer-less streaming decompression (synchronous mode)</h2><pre>
   A ZSTD_DCtx object is required to track streaming operations.
   Use ZSTD_createDCtx() / ZSTD_freeDCtx() to manage it.
-  A ZSTD_DCtx object can be re-used multiple times.
+  A ZSTD_DCtx object can be reused multiple times.
 
   First typical operation is to retrieve frame parameters, using ZSTD_getFrameHeader().
   Frame header is extracted from the beginning of compressed frame, so providing only the frame's beginning is enough.
@@ -1993,36 +2115,25 @@ ZSTDLIB_STATIC_API size_t ZSTD_compressBegin_usingCDict(ZSTD_CCtx* cctx, const Z
   For skippable frames ZSTD_decompressContinue() always returns 0 : it only skips the content.
 <BR></pre>
 
-<h3>Buffer-less streaming decompression functions</h3><pre></pre><b><pre>typedef enum { ZSTD_frame, ZSTD_skippableFrame } ZSTD_frameType_e;
-typedef struct {
-    unsigned long long frameContentSize; </b>/* if == ZSTD_CONTENTSIZE_UNKNOWN, it means this field is not available. 0 means "empty" */<b>
-    unsigned long long windowSize;       </b>/* can be very large, up to <= frameContentSize */<b>
-    unsigned blockSizeMax;
-    ZSTD_frameType_e frameType;          </b>/* if == ZSTD_skippableFrame, frameContentSize is the size of skippable content */<b>
-    unsigned headerSize;
-    unsigned dictID;
-    unsigned checksumFlag;
-    unsigned _reserved1;
-    unsigned _reserved2;
-} ZSTD_frameHeader;
-</pre></b><BR>
-<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_getFrameHeader(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize);   </b>/**< doesn't consume input */<b>
-</b>/*! ZSTD_getFrameHeader_advanced() :<b>
- *  same as ZSTD_getFrameHeader(),
- *  with added capability to select a format (like ZSTD_f_zstd1_magicless) */
-ZSTDLIB_STATIC_API size_t ZSTD_getFrameHeader_advanced(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize, ZSTD_format_e format);
-ZSTDLIB_STATIC_API size_t ZSTD_decodingBufferSize_min(unsigned long long windowSize, unsigned long long frameContentSize);  </b>/**< when frame content size is not known, pass in frameContentSize == ZSTD_CONTENTSIZE_UNKNOWN */<b>
-</b><p>  decode Frame Header, or requires larger `srcSize`.
- @return : 0, `zfhPtr` is correctly filled,
-          >0, `srcSize` is too small, value is wanted `srcSize` amount,
-           or an error code, which can be tested using ZSTD_isError() 
-</p></pre><BR>
-
+<h3>Buffer-less streaming decompression functions</h3><pre></pre><b><pre></pre></b><BR>
+<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_decodingBufferSize_min(unsigned long long windowSize, unsigned long long frameContentSize);  </b>/**< when frame content size is not known, pass in frameContentSize == ZSTD_CONTENTSIZE_UNKNOWN */<b>
+</b></pre><BR>
 <pre><b>typedef enum { ZSTDnit_frameHeader, ZSTDnit_blockHeader, ZSTDnit_block, ZSTDnit_lastBlock, ZSTDnit_checksum, ZSTDnit_skippableFrame } ZSTD_nextInputType_e;
 </b></pre><BR>
-<a name="Chapter23"></a><h2>Block level API</h2><pre></pre>
+<a name="Chapter23"></a><h2>Block level API (DEPRECATED)</h2><pre></pre>
 
-<pre><b></b><p>    Frame metadata cost is typically ~12 bytes, which can be non-negligible for very small blocks (< 100 bytes).
+<pre><b></b><p>    You can get the frame header down to 2 bytes by setting:
+      - ZSTD_c_format = ZSTD_f_zstd1_magicless
+      - ZSTD_c_contentSizeFlag = 0
+      - ZSTD_c_checksumFlag = 0
+      - ZSTD_c_dictIDFlag = 0
+
+    This API is not as well tested as our normal API, so we recommend not using it.
+    We will be removing it in a future version. If the normal API doesn't provide
+    the functionality you need, please open a GitHub issue.
+
+    Block functions produce and decode raw zstd blocks, without frame metadata.
+    Frame metadata cost is typically ~12 bytes, which can be non-negligible for very small blocks (< 100 bytes).
     But users will have to take in charge needed metadata to regenerate data, such as compressed and content sizes.
 
     A few rules to respect :
@@ -2046,36 +2157,14 @@ ZSTDLIB_STATIC_API size_t ZSTD_decodingBufferSize_min(unsigned long long windowS
         Use ZSTD_insertBlock() for such a case.
 </p></pre><BR>
 
-<h3>Raw zstd block functions</h3><pre></pre><b><pre>ZSTDLIB_STATIC_API size_t ZSTD_getBlockSize   (const ZSTD_CCtx* cctx);
+<h3>Raw zstd block functions</h3><pre></pre><b><pre>ZSTD_DEPRECATED("The block API is deprecated in favor of the normal compression API. See docs.")
+ZSTDLIB_STATIC_API size_t ZSTD_getBlockSize   (const ZSTD_CCtx* cctx);
+ZSTD_DEPRECATED("The block API is deprecated in favor of the normal compression API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_compressBlock  (ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+ZSTD_DEPRECATED("The block API is deprecated in favor of the normal compression API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_decompressBlock(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+ZSTD_DEPRECATED("The block API is deprecated in favor of the normal compression API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_insertBlock    (ZSTD_DCtx* dctx, const void* blockStart, size_t blockSize);  </b>/**< insert uncompressed block into `dctx` history. Useful for multi-blocks decompression. */<b>
 </pre></b><BR>
-<pre><b>ZSTDLIB_STATIC_API void
-ZSTD_registerSequenceProducer(
-  ZSTD_CCtx* cctx,
-  void* sequenceProducerState,
-  ZSTD_sequenceProducer_F* sequenceProducer
-);
-</b><p> Instruct zstd to use a block-level external sequence producer function.
-
- The sequenceProducerState must be initialized by the caller, and the caller is
- responsible for managing its lifetime. This parameter is sticky across
- compressions. It will remain set until the user explicitly resets compression
- parameters.
-
- Sequence producer registration is considered to be an "advanced parameter",
- part of the "advanced API". This means it will only have an effect on compression
- APIs which respect advanced parameters, such as compress2() and compressStream2().
- Older compression APIs such as compressCCtx(), which predate the introduction of
- "advanced parameters", will ignore any external sequence producer setting.
-
- The sequence producer can be "cleared" by registering a NULL function pointer. This
- removes all limitations described above in the "LIMITATIONS" section of the API docs.
-
- The user is strongly encouraged to read the full API documentation (above) before
- calling this function. 
-</p></pre><BR>
-
 </html>
 </body>

examples/streaming_compression.c

@@ -42,7 +42,13 @@ static void compressFile_orDie(const char* fname, const char* outName, int cLeve
      */
     CHECK_ZSTD( ZSTD_CCtx_setParameter(cctx, ZSTD_c_compressionLevel, cLevel) );
     CHECK_ZSTD( ZSTD_CCtx_setParameter(cctx, ZSTD_c_checksumFlag, 1) );
-    ZSTD_CCtx_setParameter(cctx, ZSTD_c_nbWorkers, nbThreads);
+    if (nbThreads > 1) {
+        size_t const r = ZSTD_CCtx_setParameter(cctx, ZSTD_c_nbWorkers, nbThreads);
+        if (ZSTD_isError(r)) {
+            fprintf (stderr, "Note: the linked libzstd library doesn't support multithreading. "
+                             "Reverting to single-thread mode. \n");
+        }
+    }
 
     /* This loop read from the input file, compresses that entire chunk,
      * and writes all output produced to the output file.
@@ -117,7 +123,7 @@ int main(int argc, const char** argv)
     }
 
     int cLevel = 1;
-    int nbThreads = 4;
+    int nbThreads = 1;
 
     if (argc >= 3) {
       cLevel = atoi (argv[2]);

lib/Makefile

@@ -8,6 +8,9 @@
 # You may select, at your option, one of the above-listed licenses.
 # ################################################################
 
+# default target (when running `make` with no argument)
+lib-release:
+
 # Modules
 ZSTD_LIB_COMPRESSION ?= 1
 ZSTD_LIB_DECOMPRESSION ?= 1
@@ -54,12 +57,11 @@ VERSION := $(ZSTD_VERSION)
 # Note: by default, the static library is built single-threaded and dynamic library is built
 # multi-threaded. It is possible to force multi or single threaded builds by appending
 # -mt or -nomt to the build target (like lib-mt for multi-threaded, lib-nomt for single-threaded).
-.PHONY: default
-default: lib-release
+
 
 CPPFLAGS_DYNLIB  += -DZSTD_MULTITHREAD # dynamic library build defaults to multi-threaded
 LDFLAGS_DYNLIB   += -pthread
-CPPFLAGS_STATLIB +=                    # static library build defaults to single-threaded
+CPPFLAGS_STATICLIB +=                  # static library build defaults to single-threaded
 
 
 ifeq ($(findstring GCC,$(CCVER)),GCC)
@@ -91,7 +93,7 @@ all: lib
 
 
 .PHONY: libzstd.a  # must be run every time
-libzstd.a: CPPFLAGS += $(CPPFLAGS_STATLIB)
+libzstd.a: CPPFLAGS += $(CPPFLAGS_STATICLIB)
 
 SET_CACHE_DIRECTORY = \
    +$(MAKE) --no-print-directory $@ \
@@ -109,19 +111,19 @@ libzstd.a:
 else
 # BUILD_DIR is defined
 
-ZSTD_STATLIB_DIR := $(BUILD_DIR)/static
-ZSTD_STATLIB := $(ZSTD_STATLIB_DIR)/libzstd.a
-ZSTD_STATLIB_OBJ := $(addprefix $(ZSTD_STATLIB_DIR)/,$(ZSTD_LOCAL_OBJ))
-$(ZSTD_STATLIB): ARFLAGS = rcs
-$(ZSTD_STATLIB): | $(ZSTD_STATLIB_DIR)
-$(ZSTD_STATLIB): $(ZSTD_STATLIB_OBJ)
+ZSTD_STATICLIB_DIR := $(BUILD_DIR)/static
+ZSTD_STATICLIB := $(ZSTD_STATICLIB_DIR)/libzstd.a
+ZSTD_STATICLIB_OBJ := $(addprefix $(ZSTD_STATICLIB_DIR)/,$(ZSTD_LOCAL_OBJ))
+$(ZSTD_STATICLIB): ARFLAGS = rcs
+$(ZSTD_STATICLIB): | $(ZSTD_STATICLIB_DIR)
+$(ZSTD_STATICLIB): $(ZSTD_STATICLIB_OBJ)
   # Check for multithread flag at target execution time
 	$(if $(filter -DZSTD_MULTITHREAD,$(CPPFLAGS)),\
     @echo compiling multi-threaded static library $(LIBVER),\
     @echo compiling single-threaded static library $(LIBVER))
 	$(AR) $(ARFLAGS) $@ $^
 
-libzstd.a: $(ZSTD_STATLIB)
+libzstd.a: $(ZSTD_STATICLIB)
 	cp -f $< $@
 
 endif
@@ -182,14 +184,14 @@ lib : libzstd.a libzstd
 # make does not consider implicit pattern rule for .PHONY target
 
 %-mt : CPPFLAGS_DYNLIB  := -DZSTD_MULTITHREAD
-%-mt : CPPFLAGS_STATLIB := -DZSTD_MULTITHREAD
+%-mt : CPPFLAGS_STATICLIB := -DZSTD_MULTITHREAD
 %-mt : LDFLAGS_DYNLIB   := -pthread
 %-mt : %
 	@echo multi-threaded build completed
 
 %-nomt : CPPFLAGS_DYNLIB  :=
 %-nomt : LDFLAGS_DYNLIB   :=
-%-nomt : CPPFLAGS_STATLIB :=
+%-nomt : CPPFLAGS_STATICLIB :=
 %-nomt : %
 	@echo single-threaded build completed
 
@@ -200,42 +202,52 @@ lib : libzstd.a libzstd
 
 # Generate .h dependencies automatically
 
-DEPFLAGS = -MT $@ -MMD -MP -MF
+# -MMD: compiler generates dependency information as a side-effect of compilation, without system headers
+# -MP: adds phony target for each dependency other than main file.
+DEPFLAGS = -MMD -MP
 
-$(ZSTD_DYNLIB_DIR)/%.o : %.c $(ZSTD_DYNLIB_DIR)/%.d | $(ZSTD_DYNLIB_DIR)
+# ensure that ZSTD_DYNLIB_DIR exists prior to generating %.o
+$(ZSTD_DYNLIB_DIR)/%.o : %.c | $(ZSTD_DYNLIB_DIR)
 	@echo CC $@
-	$(COMPILE.c) $(DEPFLAGS) $(ZSTD_DYNLIB_DIR)/$*.d $(OUTPUT_OPTION) $<
+	$(COMPILE.c) $(DEPFLAGS) $(OUTPUT_OPTION) $<
 
-$(ZSTD_STATLIB_DIR)/%.o : %.c $(ZSTD_STATLIB_DIR)/%.d | $(ZSTD_STATLIB_DIR)
+$(ZSTD_STATICLIB_DIR)/%.o : %.c | $(ZSTD_STATICLIB_DIR)
 	@echo CC $@
-	$(COMPILE.c) $(DEPFLAGS) $(ZSTD_STATLIB_DIR)/$*.d $(OUTPUT_OPTION) $<
+	$(COMPILE.c) $(DEPFLAGS) $(OUTPUT_OPTION) $<
 
 $(ZSTD_DYNLIB_DIR)/%.o : %.S | $(ZSTD_DYNLIB_DIR)
 	@echo AS $@
 	$(COMPILE.S) $(OUTPUT_OPTION) $<
 
-$(ZSTD_STATLIB_DIR)/%.o : %.S | $(ZSTD_STATLIB_DIR)
+$(ZSTD_STATICLIB_DIR)/%.o : %.S | $(ZSTD_STATICLIB_DIR)
 	@echo AS $@
 	$(COMPILE.S) $(OUTPUT_OPTION) $<
 
-MKDIR ?= mkdir
-$(BUILD_DIR) $(ZSTD_DYNLIB_DIR) $(ZSTD_STATLIB_DIR):
-	$(MKDIR) -p $@
+MKDIR ?= mkdir -p
+$(BUILD_DIR) $(ZSTD_DYNLIB_DIR) $(ZSTD_STATICLIB_DIR):
+	$(MKDIR) $@
 
-DEPFILES := $(ZSTD_DYNLIB_OBJ:.o=.d) $(ZSTD_STATLIB_OBJ:.o=.d)
+DEPFILES := $(ZSTD_DYNLIB_OBJ:.o=.d) $(ZSTD_STATICLIB_OBJ:.o=.d)
 $(DEPFILES):
 
-include $(wildcard $(DEPFILES))
+# The leading '-' means: do not fail is include fails (ex: directory does not exist yet)
+-include $(wildcard $(DEPFILES))
 
 
-# Special case : building library in single-thread mode _and_ without zstdmt_compress.c
-ZSTDMT_FILES = compress/zstdmt_compress.c
-ZSTD_NOMT_FILES = $(filter-out $(ZSTDMT_FILES),$(ZSTD_FILES))
+# Special case : build library in single-thread mode _and_ without zstdmt_compress.c
+# Note : we still need threading.c and pool.c for the dictionary builder,
+# but they will correctly behave single-threaded.
+ZSTDMT_FILES = zstdmt_compress.c
+ZSTD_NOMT_FILES = $(filter-out $(ZSTDMT_FILES),$(notdir $(ZSTD_FILES)))
 libzstd-nomt: CFLAGS += -fPIC -fvisibility=hidden
 libzstd-nomt: LDFLAGS += -shared
 libzstd-nomt: $(ZSTD_NOMT_FILES)
 	@echo compiling single-thread dynamic library $(LIBVER)
 	@echo files : $(ZSTD_NOMT_FILES)
+	@if echo "$(ZSTD_NOMT_FILES)" | tr ' ' '\n' | $(GREP) -q zstdmt; then \
+        echo "Error: Found zstdmt in list."; \
+        exit 1; \
+    fi
 	$(CC) $(FLAGS) $^ $(LDFLAGS) $(SONAME_FLAGS) -o $@
 
 .PHONY: clean
@@ -249,7 +261,7 @@ clean:
 #-----------------------------------------------------------------------------
 # make install is validated only for below listed environments
 #-----------------------------------------------------------------------------
-ifneq (,$(filter $(UNAME),Linux Darwin GNU/kFreeBSD GNU OpenBSD FreeBSD NetBSD DragonFly SunOS Haiku AIX))
+ifneq (,$(filter $(UNAME),Linux Darwin GNU/kFreeBSD GNU OpenBSD FreeBSD NetBSD DragonFly SunOS Haiku AIX MSYS_NT CYGWIN_NT))
 
 lib: libzstd.pc
 

lib/README.md

@@ -88,7 +88,7 @@ The file structure is designed to make this selection manually achievable for an
         For example, advanced API for version `v0.4` is exposed in `lib/legacy/zstd_v04.h` .
 
 - While invoking `make libzstd`, it's possible to define build macros
-        `ZSTD_LIB_COMPRESSION, ZSTD_LIB_DECOMPRESSION`, `ZSTD_LIB_DICTBUILDER`,
+        `ZSTD_LIB_COMPRESSION`, `ZSTD_LIB_DECOMPRESSION`, `ZSTD_LIB_DICTBUILDER`,
         and `ZSTD_LIB_DEPRECATED` as `0` to forgo compilation of the
         corresponding features. This will also disable compilation of all
         dependencies (e.g. `ZSTD_LIB_COMPRESSION=0` will also disable
@@ -119,6 +119,15 @@ The file structure is designed to make this selection manually achievable for an
   binary is achieved by using `HUF_FORCE_DECOMPRESS_X1` and
   `ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT` (implied by `ZSTD_LIB_MINIFY`).
 
+  On the compressor side, Zstd's compression levels map to several internal
+  strategies. In environments where the higher compression levels aren't used,
+  it is possible to exclude all but the fastest strategy with
+  `ZSTD_LIB_EXCLUDE_COMPRESSORS_DFAST_AND_UP=1`. (Note that this will change
+  the behavior of the default compression level.) Or if you want to retain the
+  default compressor as well, you can set
+  `ZSTD_LIB_EXCLUDE_COMPRESSORS_GREEDY_AND_UP=1`, at the cost of an additional
+  ~20KB or so.
+
   For squeezing the last ounce of size out, you can also define
   `ZSTD_NO_INLINE`, which disables inlining, and `ZSTD_STRIP_ERROR_STRINGS`,
   which removes the error messages that are otherwise returned by
@@ -169,6 +178,10 @@ The file structure is designed to make this selection manually achievable for an
   `ZSTDERRORLIB_VSIBILITY`, and `ZDICTLIB_VISIBILITY` if unset, for backwards compatibility
   with the old macro names.
 
+- The C compiler macro `HUF_DISABLE_FAST_DECODE` disables the newer Huffman fast C
+  and assembly decoding loops. You may want to use this macro if these loops are
+  slower on your platform.
+
 #### Windows : using MinGW+MSYS to create DLL
 
 DLL can be created using MinGW+MSYS with the `make libzstd` command.

lib/common/allocations.h

@@ -0,0 +1,55 @@
+/*
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+/* This file provides custom allocation primitives
+ */
+
+#define ZSTD_DEPS_NEED_MALLOC
+#include "zstd_deps.h"   /* ZSTD_malloc, ZSTD_calloc, ZSTD_free, ZSTD_memset */
+
+#include "compiler.h" /* MEM_STATIC */
+#define ZSTD_STATIC_LINKING_ONLY
+#include "../zstd.h" /* ZSTD_customMem */
+
+#ifndef ZSTD_ALLOCATIONS_H
+#define ZSTD_ALLOCATIONS_H
+
+/* custom memory allocation functions */
+
+MEM_STATIC void* ZSTD_customMalloc(size_t size, ZSTD_customMem customMem)
+{
+    if (customMem.customAlloc)
+        return customMem.customAlloc(customMem.opaque, size);
+    return ZSTD_malloc(size);
+}
+
+MEM_STATIC void* ZSTD_customCalloc(size_t size, ZSTD_customMem customMem)
+{
+    if (customMem.customAlloc) {
+        /* calloc implemented as malloc+memset;
+         * not as efficient as calloc, but next best guess for custom malloc */
+        void* const ptr = customMem.customAlloc(customMem.opaque, size);
+        ZSTD_memset(ptr, 0, size);
+        return ptr;
+    }
+    return ZSTD_calloc(1, size);
+}
+
+MEM_STATIC void ZSTD_customFree(void* ptr, ZSTD_customMem customMem)
+{
+    if (ptr!=NULL) {
+        if (customMem.customFree)
+            customMem.customFree(customMem.opaque, ptr);
+        else
+            ZSTD_free(ptr);
+    }
+}
+
+#endif /* ZSTD_ALLOCATIONS_H */

lib/common/bits.h

@@ -17,7 +17,7 @@ MEM_STATIC unsigned ZSTD_countTrailingZeros32_fallback(U32 val)
 {
     assert(val != 0);
     {
-        static const int DeBruijnBytePos[32] = {0, 1, 28, 2, 29, 14, 24, 3,
+        static const U32 DeBruijnBytePos[32] = {0, 1, 28, 2, 29, 14, 24, 3,
                                                 30, 22, 20, 15, 25, 17, 4, 8,
                                                 31, 27, 13, 23, 21, 19, 16, 7,
                                                 26, 12, 18, 6, 11, 5, 10, 9};
@@ -30,7 +30,7 @@ MEM_STATIC unsigned ZSTD_countTrailingZeros32(U32 val)
     assert(val != 0);
 #   if defined(_MSC_VER)
 #       if STATIC_BMI2 == 1
-            return _tzcnt_u32(val);
+            return (unsigned)_tzcnt_u32(val);
 #       else
             if (val != 0) {
                 unsigned long r;
@@ -69,7 +69,7 @@ MEM_STATIC unsigned ZSTD_countLeadingZeros32(U32 val)
     assert(val != 0);
 #   if defined(_MSC_VER)
 #       if STATIC_BMI2 == 1
-            return _lzcnt_u32(val);
+            return (unsigned)_lzcnt_u32(val);
 #       else
             if (val != 0) {
                 unsigned long r;
@@ -92,7 +92,7 @@ MEM_STATIC unsigned ZSTD_countTrailingZeros64(U64 val)
     assert(val != 0);
 #   if defined(_MSC_VER) && defined(_WIN64)
 #       if STATIC_BMI2 == 1
-            return _tzcnt_u64(val);
+            return (unsigned)_tzcnt_u64(val);
 #       else
             if (val != 0) {
                 unsigned long r;
@@ -123,7 +123,7 @@ MEM_STATIC unsigned ZSTD_countLeadingZeros64(U64 val)
     assert(val != 0);
 #   if defined(_MSC_VER) && defined(_WIN64)
 #       if STATIC_BMI2 == 1
-            return _lzcnt_u64(val);
+            return (unsigned)_lzcnt_u64(val);
 #       else
             if (val != 0) {
                 unsigned long r;
@@ -172,4 +172,29 @@ MEM_STATIC unsigned ZSTD_highbit32(U32 val)   /* compress, dictBuilder, decodeCo
     return 31 - ZSTD_countLeadingZeros32(val);
 }
 
+/* ZSTD_rotateRight_*():
+ * Rotates a bitfield to the right by "count" bits.
+ * https://en.wikipedia.org/w/index.php?title=Circular_shift&oldid=991635599#Implementing_circular_shifts
+ */
+MEM_STATIC
+U64 ZSTD_rotateRight_U64(U64 const value, U32 count) {
+    assert(count < 64);
+    count &= 0x3F; /* for fickle pattern recognition */
+    return (value >> count) | (U64)(value << ((0U - count) & 0x3F));
+}
+
+MEM_STATIC
+U32 ZSTD_rotateRight_U32(U32 const value, U32 count) {
+    assert(count < 32);
+    count &= 0x1F; /* for fickle pattern recognition */
+    return (value >> count) | (U32)(value << ((0U - count) & 0x1F));
+}
+
+MEM_STATIC
+U16 ZSTD_rotateRight_U16(U16 const value, U32 count) {
+    assert(count < 16);
+    count &= 0x0F; /* for fickle pattern recognition */
+    return (value >> count) | (U16)(value << ((0U - count) & 0x0F));
+}
+
 #endif /* ZSTD_BITS_H */

lib/common/bitstream.h

@@ -90,19 +90,20 @@ MEM_STATIC size_t BIT_closeCStream(BIT_CStream_t* bitC);
 /*-********************************************
 *  bitStream decoding API (read backward)
 **********************************************/
+typedef size_t BitContainerType;
 typedef struct {
-    size_t   bitContainer;
+    BitContainerType bitContainer;
     unsigned bitsConsumed;
     const char* ptr;
     const char* start;
     const char* limitPtr;
 } BIT_DStream_t;
 
-typedef enum { BIT_DStream_unfinished = 0,
-               BIT_DStream_endOfBuffer = 1,
-               BIT_DStream_completed = 2,
-               BIT_DStream_overflow = 3 } BIT_DStream_status;  /* result of BIT_reloadDStream() */
-               /* 1,2,4,8 would be better for bitmap combinations, but slows down performance a bit ... :( */
+typedef enum { BIT_DStream_unfinished = 0,  /* fully refilled */
+               BIT_DStream_endOfBuffer = 1, /* still some bits left in bitstream */
+               BIT_DStream_completed = 2,   /* bitstream entirely consumed, bit-exact */
+               BIT_DStream_overflow = 3     /* user requested more bits than present in bitstream */
+    } BIT_DStream_status;  /* result of BIT_reloadDStream() */
 
 MEM_STATIC size_t   BIT_initDStream(BIT_DStream_t* bitD, const void* srcBuffer, size_t srcSize);
 MEM_STATIC size_t   BIT_readBits(BIT_DStream_t* bitD, unsigned nbBits);
@@ -112,7 +113,7 @@ MEM_STATIC unsigned BIT_endOfDStream(const BIT_DStream_t* bitD);
 
 /* Start by invoking BIT_initDStream().
 *  A chunk of the bitStream is then stored into a local register.
-*  Local register size is 64-bits on 64-bits systems, 32-bits on 32-bits systems (size_t).
+*  Local register size is 64-bits on 64-bits systems, 32-bits on 32-bits systems (BitContainerType).
 *  You can then retrieve bitFields stored into the local register, **in reverse order**.
 *  Local register is explicitly reloaded from memory by the BIT_reloadDStream() method.
 *  A reload guarantee a minimum of ((8*sizeof(bitD->bitContainer))-7) bits when its result is BIT_DStream_unfinished.
@@ -162,7 +163,7 @@ MEM_STATIC size_t BIT_initCStream(BIT_CStream_t* bitC,
     return 0;
 }
 
-MEM_STATIC FORCE_INLINE_ATTR size_t BIT_getLowerBits(size_t bitContainer, U32 const nbBits)
+FORCE_INLINE_TEMPLATE size_t BIT_getLowerBits(size_t bitContainer, U32 const nbBits)
 {
 #if defined(STATIC_BMI2) && STATIC_BMI2 == 1 && !defined(ZSTD_NO_INTRINSICS)
     return  _bzhi_u64(bitContainer, nbBits);
@@ -267,22 +268,22 @@ MEM_STATIC size_t BIT_initDStream(BIT_DStream_t* bitD, const void* srcBuffer, si
         bitD->bitContainer = *(const BYTE*)(bitD->start);
         switch(srcSize)
         {
-        case 7: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[6]) << (sizeof(bitD->bitContainer)*8 - 16);
+        case 7: bitD->bitContainer += (BitContainerType)(((const BYTE*)(srcBuffer))[6]) << (sizeof(bitD->bitContainer)*8 - 16);
                 ZSTD_FALLTHROUGH;
 
-        case 6: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[5]) << (sizeof(bitD->bitContainer)*8 - 24);
+        case 6: bitD->bitContainer += (BitContainerType)(((const BYTE*)(srcBuffer))[5]) << (sizeof(bitD->bitContainer)*8 - 24);
                 ZSTD_FALLTHROUGH;
 
-        case 5: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[4]) << (sizeof(bitD->bitContainer)*8 - 32);
+        case 5: bitD->bitContainer += (BitContainerType)(((const BYTE*)(srcBuffer))[4]) << (sizeof(bitD->bitContainer)*8 - 32);
                 ZSTD_FALLTHROUGH;
 
-        case 4: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[3]) << 24;
+        case 4: bitD->bitContainer += (BitContainerType)(((const BYTE*)(srcBuffer))[3]) << 24;
                 ZSTD_FALLTHROUGH;
 
-        case 3: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[2]) << 16;
+        case 3: bitD->bitContainer += (BitContainerType)(((const BYTE*)(srcBuffer))[2]) << 16;
                 ZSTD_FALLTHROUGH;
 
-        case 2: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[1]) <<  8;
+        case 2: bitD->bitContainer += (BitContainerType)(((const BYTE*)(srcBuffer))[1]) <<  8;
                 ZSTD_FALLTHROUGH;
 
         default: break;
@@ -297,12 +298,12 @@ MEM_STATIC size_t BIT_initDStream(BIT_DStream_t* bitD, const void* srcBuffer, si
     return srcSize;
 }
 
-MEM_STATIC FORCE_INLINE_ATTR size_t BIT_getUpperBits(size_t bitContainer, U32 const start)
+FORCE_INLINE_TEMPLATE size_t BIT_getUpperBits(BitContainerType bitContainer, U32 const start)
 {
     return bitContainer >> start;
 }
 
-MEM_STATIC FORCE_INLINE_ATTR size_t BIT_getMiddleBits(size_t bitContainer, U32 const start, U32 const nbBits)
+FORCE_INLINE_TEMPLATE size_t BIT_getMiddleBits(BitContainerType bitContainer, U32 const start, U32 const nbBits)
 {
     U32 const regMask = sizeof(bitContainer)*8 - 1;
     /* if start > regMask, bitstream is corrupted, and result is undefined */
@@ -325,7 +326,7 @@ MEM_STATIC FORCE_INLINE_ATTR size_t BIT_getMiddleBits(size_t bitContainer, U32 c
  *  On 32-bits, maxNbBits==24.
  *  On 64-bits, maxNbBits==56.
  * @return : value extracted */
-MEM_STATIC  FORCE_INLINE_ATTR size_t BIT_lookBits(const BIT_DStream_t*  bitD, U32 nbBits)
+FORCE_INLINE_TEMPLATE size_t BIT_lookBits(const BIT_DStream_t*  bitD, U32 nbBits)
 {
     /* arbitrate between double-shift and shift+mask */
 #if 1
@@ -348,7 +349,7 @@ MEM_STATIC size_t BIT_lookBitsFast(const BIT_DStream_t* bitD, U32 nbBits)
     return (bitD->bitContainer << (bitD->bitsConsumed & regMask)) >> (((regMask+1)-nbBits) & regMask);
 }
 
-MEM_STATIC FORCE_INLINE_ATTR void BIT_skipBits(BIT_DStream_t* bitD, U32 nbBits)
+FORCE_INLINE_TEMPLATE void BIT_skipBits(BIT_DStream_t* bitD, U32 nbBits)
 {
     bitD->bitsConsumed += nbBits;
 }
@@ -357,7 +358,7 @@ MEM_STATIC FORCE_INLINE_ATTR void BIT_skipBits(BIT_DStream_t* bitD, U32 nbBits)
  *  Read (consume) next n bits from local register and update.
  *  Pay attention to not read more than nbBits contained into local register.
  * @return : extracted value. */
-MEM_STATIC FORCE_INLINE_ATTR size_t BIT_readBits(BIT_DStream_t* bitD, unsigned nbBits)
+FORCE_INLINE_TEMPLATE size_t BIT_readBits(BIT_DStream_t* bitD, unsigned nbBits)
 {
     size_t const value = BIT_lookBits(bitD, nbBits);
     BIT_skipBits(bitD, nbBits);
@@ -374,6 +375,21 @@ MEM_STATIC size_t BIT_readBitsFast(BIT_DStream_t* bitD, unsigned nbBits)
     return value;
 }
 
+/*! BIT_reloadDStream_internal() :
+ *  Simple variant of BIT_reloadDStream(), with two conditions:
+ *  1. bitstream is valid : bitsConsumed <= sizeof(bitD->bitContainer)*8
+ *  2. look window is valid after shifted down : bitD->ptr >= bitD->start
+ */
+MEM_STATIC BIT_DStream_status BIT_reloadDStream_internal(BIT_DStream_t* bitD)
+{
+    assert(bitD->bitsConsumed <= sizeof(bitD->bitContainer)*8);
+    bitD->ptr -= bitD->bitsConsumed >> 3;
+    assert(bitD->ptr >= bitD->start);
+    bitD->bitsConsumed &= 7;
+    bitD->bitContainer = MEM_readLEST(bitD->ptr);
+    return BIT_DStream_unfinished;
+}
+
 /*! BIT_reloadDStreamFast() :
  *  Similar to BIT_reloadDStream(), but with two differences:
  *  1. bitsConsumed <= sizeof(bitD->bitContainer)*8 must hold!
@@ -384,31 +400,35 @@ MEM_STATIC BIT_DStream_status BIT_reloadDStreamFast(BIT_DStream_t* bitD)
 {
     if (UNLIKELY(bitD->ptr < bitD->limitPtr))
         return BIT_DStream_overflow;
-    assert(bitD->bitsConsumed <= sizeof(bitD->bitContainer)*8);
-    bitD->ptr -= bitD->bitsConsumed >> 3;
-    bitD->bitsConsumed &= 7;
-    bitD->bitContainer = MEM_readLEST(bitD->ptr);
-    return BIT_DStream_unfinished;
+    return BIT_reloadDStream_internal(bitD);
 }
 
 /*! BIT_reloadDStream() :
  *  Refill `bitD` from buffer previously set in BIT_initDStream() .
- *  This function is safe, it guarantees it will not read beyond src buffer.
+ *  This function is safe, it guarantees it will not never beyond src buffer.
  * @return : status of `BIT_DStream_t` internal register.
  *           when status == BIT_DStream_unfinished, internal register is filled with at least 25 or 57 bits */
-MEM_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t* bitD)
+FORCE_INLINE_TEMPLATE BIT_DStream_status BIT_reloadDStream(BIT_DStream_t* bitD)
 {
-    if (bitD->bitsConsumed > (sizeof(bitD->bitContainer)*8))  /* overflow detected, like end of stream */
+    /* note : once in overflow mode, a bitstream remains in this mode until it's reset */
+    if (UNLIKELY(bitD->bitsConsumed > (sizeof(bitD->bitContainer)*8))) {
+        static const BitContainerType zeroFilled = 0;
+        bitD->ptr = (const char*)&zeroFilled; /* aliasing is allowed for char */
+        /* overflow detected, erroneous scenario or end of stream: no update */
         return BIT_DStream_overflow;
+    }
+
+    assert(bitD->ptr >= bitD->start);
 
     if (bitD->ptr >= bitD->limitPtr) {
-        return BIT_reloadDStreamFast(bitD);
+        return BIT_reloadDStream_internal(bitD);
     }
     if (bitD->ptr == bitD->start) {
+        /* reached end of bitStream => no update */
         if (bitD->bitsConsumed < sizeof(bitD->bitContainer)*8) return BIT_DStream_endOfBuffer;
         return BIT_DStream_completed;
     }
-    /* start < ptr < limitPtr */
+    /* start < ptr < limitPtr => cautious update */
     {   U32 nbBytes = bitD->bitsConsumed >> 3;
         BIT_DStream_status result = BIT_DStream_unfinished;
         if (bitD->ptr - nbBytes < bitD->start) {

lib/common/compiler.h

@@ -11,6 +11,8 @@
 #ifndef ZSTD_COMPILER_H
 #define ZSTD_COMPILER_H
 
+#include <stddef.h>
+
 #include "portability_macros.h"
 
 /*-*******************************************************
@@ -51,12 +53,19 @@
 #  define WIN_CDECL
 #endif
 
+/* UNUSED_ATTR tells the compiler it is okay if the function is unused. */
+#if defined(__GNUC__)
+#  define UNUSED_ATTR __attribute__((unused))
+#else
+#  define UNUSED_ATTR
+#endif
+
 /**
  * FORCE_INLINE_TEMPLATE is used to define C "templates", which take constant
  * parameters. They must be inlined for the compiler to eliminate the constant
  * branches.
  */
-#define FORCE_INLINE_TEMPLATE static INLINE_KEYWORD FORCE_INLINE_ATTR
+#define FORCE_INLINE_TEMPLATE static INLINE_KEYWORD FORCE_INLINE_ATTR UNUSED_ATTR
 /**
  * HINT_INLINE is used to help the compiler generate better code. It is *not*
  * used for "templates", so it can be tweaked based on the compilers
@@ -71,14 +80,28 @@
 #if !defined(__clang__) && defined(__GNUC__) && __GNUC__ >= 4 && __GNUC_MINOR__ >= 8 && __GNUC__ < 5
 #  define HINT_INLINE static INLINE_KEYWORD
 #else
-#  define HINT_INLINE static INLINE_KEYWORD FORCE_INLINE_ATTR
+#  define HINT_INLINE FORCE_INLINE_TEMPLATE
 #endif
 
-/* UNUSED_ATTR tells the compiler it is okay if the function is unused. */
+/* "soft" inline :
+ * The compiler is free to select if it's a good idea to inline or not.
+ * The main objective is to silence compiler warnings
+ * when a defined function in included but not used.
+ *
+ * Note : this macro is prefixed `MEM_` because it used to be provided by `mem.h` unit.
+ * Updating the prefix is probably preferable, but requires a fairly large codemod,
+ * since this name is used everywhere.
+ */
+#ifndef MEM_STATIC  /* already defined in Linux Kernel mem.h */
 #if defined(__GNUC__)
-#  define UNUSED_ATTR __attribute__((unused))
+#  define MEM_STATIC static __inline UNUSED_ATTR
+#elif defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */)
+#  define MEM_STATIC static inline
+#elif defined(_MSC_VER)
+#  define MEM_STATIC static __inline
 #else
-#  define UNUSED_ATTR
+#  define MEM_STATIC static  /* this version may generate warnings for unused static functions; disable the relevant warning */
+#endif
 #endif
 
 /* force no inlining */
@@ -109,10 +132,10 @@
 /* prefetch
  * can be disabled, by declaring NO_PREFETCH build macro */
 #if defined(NO_PREFETCH)
-#  define PREFETCH_L1(ptr)  (void)(ptr)  /* disabled */
-#  define PREFETCH_L2(ptr)  (void)(ptr)  /* disabled */
+#  define PREFETCH_L1(ptr)  do { (void)(ptr); } while (0)  /* disabled */
+#  define PREFETCH_L2(ptr)  do { (void)(ptr); } while (0)  /* disabled */
 #else
-#  if defined(_MSC_VER) && (defined(_M_X64) || defined(_M_I86))  /* _mm_prefetch() is not defined outside of x86/x64 */
+#  if defined(_MSC_VER) && (defined(_M_X64) || defined(_M_I86)) && !defined(_M_ARM64EC)  /* _mm_prefetch() is not defined outside of x86/x64 */
 #    include <mmintrin.h>   /* https://msdn.microsoft.com/fr-fr/library/84szxsww(v=vs.90).aspx */
 #    define PREFETCH_L1(ptr)  _mm_prefetch((const char*)(ptr), _MM_HINT_T0)
 #    define PREFETCH_L2(ptr)  _mm_prefetch((const char*)(ptr), _MM_HINT_T1)
@@ -120,24 +143,25 @@
 #    define PREFETCH_L1(ptr)  __builtin_prefetch((ptr), 0 /* rw==read */, 3 /* locality */)
 #    define PREFETCH_L2(ptr)  __builtin_prefetch((ptr), 0 /* rw==read */, 2 /* locality */)
 #  elif defined(__aarch64__)
-#    define PREFETCH_L1(ptr)  __asm__ __volatile__("prfm pldl1keep, %0" ::"Q"(*(ptr)))
-#    define PREFETCH_L2(ptr)  __asm__ __volatile__("prfm pldl2keep, %0" ::"Q"(*(ptr)))
+#    define PREFETCH_L1(ptr)  do { __asm__ __volatile__("prfm pldl1keep, %0" ::"Q"(*(ptr))); } while (0)
+#    define PREFETCH_L2(ptr)  do { __asm__ __volatile__("prfm pldl2keep, %0" ::"Q"(*(ptr))); } while (0)
 #  else
-#    define PREFETCH_L1(ptr) (void)(ptr)  /* disabled */
-#    define PREFETCH_L2(ptr) (void)(ptr)  /* disabled */
+#    define PREFETCH_L1(ptr) do { (void)(ptr); } while (0)  /* disabled */
+#    define PREFETCH_L2(ptr) do { (void)(ptr); } while (0)  /* disabled */
 #  endif
 #endif  /* NO_PREFETCH */
 
 #define CACHELINE_SIZE 64
 
-#define PREFETCH_AREA(p, s)  {            \
-    const char* const _ptr = (const char*)(p);  \
-    size_t const _size = (size_t)(s);     \
-    size_t _pos;                          \
-    for (_pos=0; _pos<_size; _pos+=CACHELINE_SIZE) {  \
-        PREFETCH_L2(_ptr + _pos);         \
-    }                                     \
-}
+#define PREFETCH_AREA(p, s)                              \
+    do {                                                 \
+        const char* const _ptr = (const char*)(p);       \
+        size_t const _size = (size_t)(s);                \
+        size_t _pos;                                     \
+        for (_pos=0; _pos<_size; _pos+=CACHELINE_SIZE) { \
+            PREFETCH_L2(_ptr + _pos);                    \
+        }                                                \
+    } while (0)
 
 /* vectorization
  * older GCC (pre gcc-4.3 picked as the cutoff) uses a different syntax,
@@ -166,9 +190,9 @@
 #endif
 
 #if __has_builtin(__builtin_unreachable) || (defined(__GNUC__) && (__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 5)))
-#  define ZSTD_UNREACHABLE { assert(0), __builtin_unreachable(); }
+#  define ZSTD_UNREACHABLE do { assert(0), __builtin_unreachable(); } while (0)
 #else
-#  define ZSTD_UNREACHABLE { assert(0); }
+#  define ZSTD_UNREACHABLE do { assert(0); } while (0)
 #endif
 
 /* disable warnings */
@@ -281,6 +305,74 @@
 *  Sanitizer
 *****************************************************************/
 
+/**
+ * Zstd relies on pointer overflow in its decompressor.
+ * We add this attribute to functions that rely on pointer overflow.
+ */
+#ifndef ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+#  if __has_attribute(no_sanitize)
+#    if !defined(__clang__) && defined(__GNUC__) && __GNUC__ < 8
+       /* gcc < 8 only has signed-integer-overlow which triggers on pointer overflow */
+#      define ZSTD_ALLOW_POINTER_OVERFLOW_ATTR __attribute__((no_sanitize("signed-integer-overflow")))
+#    else
+       /* older versions of clang [3.7, 5.0) will warn that pointer-overflow is ignored. */
+#      define ZSTD_ALLOW_POINTER_OVERFLOW_ATTR __attribute__((no_sanitize("pointer-overflow")))
+#    endif
+#  else
+#    define ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+#  endif
+#endif
+
+/**
+ * Helper function to perform a wrapped pointer difference without trigging
+ * UBSAN.
+ *
+ * @returns lhs - rhs with wrapping
+ */
+MEM_STATIC
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+ptrdiff_t ZSTD_wrappedPtrDiff(unsigned char const* lhs, unsigned char const* rhs)
+{
+    return lhs - rhs;
+}
+
+/**
+ * Helper function to perform a wrapped pointer add without triggering UBSAN.
+ *
+ * @return ptr + add with wrapping
+ */
+MEM_STATIC
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+unsigned char const* ZSTD_wrappedPtrAdd(unsigned char const* ptr, ptrdiff_t add)
+{
+    return ptr + add;
+}
+
+/**
+ * Helper function to perform a wrapped pointer subtraction without triggering
+ * UBSAN.
+ *
+ * @return ptr - sub with wrapping
+ */
+MEM_STATIC
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+unsigned char const* ZSTD_wrappedPtrSub(unsigned char const* ptr, ptrdiff_t sub)
+{
+    return ptr - sub;
+}
+
+/**
+ * Helper function to add to a pointer that works around C's undefined behavior
+ * of adding 0 to NULL.
+ *
+ * @returns `ptr + add` except it defines `NULL + 0 == NULL`.
+ */
+MEM_STATIC
+unsigned char* ZSTD_maybeNullPtrAdd(unsigned char* ptr, ptrdiff_t add)
+{
+    return add > 0 ? ptr + add : ptr;
+}
+
 /* Issue #3240 reports an ASAN failure on an llvm-mingw build. Out of an
  * abundance of caution, disable our custom poisoning on mingw. */
 #ifdef __MINGW32__
@@ -311,6 +403,10 @@ void __msan_poison(const volatile void *a, size_t size);
 /* Returns the offset of the first (at least partially) poisoned byte in the
    memory range, or -1 if the whole range is good. */
 intptr_t __msan_test_shadow(const volatile void *x, size_t size);
+
+/* Print shadow and origin for the memory range to stderr in a human-readable
+   format. */
+void __msan_print_shadow(const volatile void *x, size_t size);
 #endif
 
 #if ZSTD_ADDRESS_SANITIZER && !defined(ZSTD_ASAN_DONT_POISON_WORKSPACE)

lib/common/cpu.h

@@ -35,6 +35,7 @@ MEM_STATIC ZSTD_cpuid_t ZSTD_cpuid(void) {
     U32 f7b = 0;
     U32 f7c = 0;
 #if defined(_MSC_VER) && (defined(_M_X64) || defined(_M_IX86))
+#if !defined(__clang__)
     int reg[4];
     __cpuid((int*)reg, 0);
     {
@@ -50,6 +51,41 @@ MEM_STATIC ZSTD_cpuid_t ZSTD_cpuid(void) {
             f7c = (U32)reg[2];
         }
     }
+#else
+    /* Clang compiler has a bug (fixed in https://reviews.llvm.org/D101338) in
+     * which the `__cpuid` intrinsic does not save and restore `rbx` as it needs
+     * to due to being a reserved register. So in that case, do the `cpuid`
+     * ourselves. Clang supports inline assembly anyway.
+     */
+    U32 n;
+    __asm__(
+        "pushq %%rbx\n\t"
+        "cpuid\n\t"
+        "popq %%rbx\n\t"
+        : "=a"(n)
+        : "a"(0)
+        : "rcx", "rdx");
+    if (n >= 1) {
+      U32 f1a;
+      __asm__(
+          "pushq %%rbx\n\t"
+          "cpuid\n\t"
+          "popq %%rbx\n\t"
+          : "=a"(f1a), "=c"(f1c), "=d"(f1d)
+          : "a"(1)
+          :);
+    }
+    if (n >= 7) {
+      __asm__(
+          "pushq %%rbx\n\t"
+          "cpuid\n\t"
+          "movq %%rbx, %%rax\n\t"
+          "popq %%rbx"
+          : "=a"(f7b), "=c"(f7c)
+          : "a"(7), "c"(0)
+          : "rdx");
+    }
+#endif
 #elif defined(__i386__) && defined(__PIC__) && !defined(__clang__) && defined(__GNUC__)
     /* The following block like the normal cpuid branch below, but gcc
      * reserves ebx for use of its pic register so we must specially

lib/common/debug.c

@@ -21,4 +21,10 @@
 
 #include "debug.h"
 
+#if !defined(ZSTD_LINUX_KERNEL) || (DEBUGLEVEL>=2)
+/* We only use this when DEBUGLEVEL>=2, but we get -Werror=pedantic errors if a
+ * translation unit is empty. So remove this from Linux kernel builds, but
+ * otherwise just leave it in.
+ */
 int g_debuglevel = DEBUGLEVEL;
+#endif

lib/common/debug.h

@@ -85,18 +85,27 @@ extern int g_debuglevel; /* the variable is only declared,
                             It's useful when enabling very verbose levels
                             on selective conditions (such as position in src) */
 
-#  define RAWLOG(l, ...) {                                       \
-                if (l<=g_debuglevel) {                           \
-                    ZSTD_DEBUG_PRINT(__VA_ARGS__);               \
-            }   }
-#  define DEBUGLOG(l, ...) {                                     \
-                if (l<=g_debuglevel) {                           \
-                    ZSTD_DEBUG_PRINT(__FILE__ ": " __VA_ARGS__); \
-                    ZSTD_DEBUG_PRINT(" \n");                     \
-            }   }
+#  define RAWLOG(l, ...)                   \
+    do {                                   \
+        if (l<=g_debuglevel) {             \
+            ZSTD_DEBUG_PRINT(__VA_ARGS__); \
+        }                                  \
+    } while (0)
+
+#define STRINGIFY(x) #x
+#define TOSTRING(x) STRINGIFY(x)
+#define LINE_AS_STRING TOSTRING(__LINE__)
+
+#  define DEBUGLOG(l, ...)                               \
+    do {                                                 \
+        if (l<=g_debuglevel) {                           \
+            ZSTD_DEBUG_PRINT(__FILE__ ":" LINE_AS_STRING ": " __VA_ARGS__); \
+            ZSTD_DEBUG_PRINT(" \n");                     \
+        }                                                \
+    } while (0)
 #else
-#  define RAWLOG(l, ...)      {}    /* disabled */
-#  define DEBUGLOG(l, ...)    {}    /* disabled */
+#  define RAWLOG(l, ...)   do { } while (0)    /* disabled */
+#  define DEBUGLOG(l, ...) do { } while (0)    /* disabled */
 #endif
 
 

lib/common/error_private.h

@@ -60,8 +60,13 @@ ERR_STATIC unsigned ERR_isError(size_t code) { return (code > ERROR(maxCode)); }
 ERR_STATIC ERR_enum ERR_getErrorCode(size_t code) { if (!ERR_isError(code)) return (ERR_enum)0; return (ERR_enum) (0-code); }
 
 /* check and forward error code */
-#define CHECK_V_F(e, f) size_t const e = f; if (ERR_isError(e)) return e
-#define CHECK_F(f)   { CHECK_V_F(_var_err__, f); }
+#define CHECK_V_F(e, f)     \
+    size_t const e = f;     \
+    do {                    \
+        if (ERR_isError(e)) \
+            return e;       \
+    } while (0)
+#define CHECK_F(f)   do { CHECK_V_F(_var_err__, f); } while (0)
 
 
 /*-****************************************
@@ -95,10 +100,12 @@ void _force_has_format_string(const char *format, ...) {
  * We want to force this function invocation to be syntactically correct, but
  * we don't want to force runtime evaluation of its arguments.
  */
-#define _FORCE_HAS_FORMAT_STRING(...) \
-  if (0) { \
-    _force_has_format_string(__VA_ARGS__); \
-  }
+#define _FORCE_HAS_FORMAT_STRING(...)              \
+    do {                                           \
+        if (0) {                                   \
+            _force_has_format_string(__VA_ARGS__); \
+        }                                          \
+    } while (0)
 
 #define ERR_QUOTE(str) #str
 
@@ -109,48 +116,50 @@ void _force_has_format_string(const char *format, ...) {
  * In order to do that (particularly, printing the conditional that failed),
  * this can't just wrap RETURN_ERROR().
  */
-#define RETURN_ERROR_IF(cond, err, ...) \
-  if (cond) { \
-    RAWLOG(3, "%s:%d: ERROR!: check %s failed, returning %s", \
-           __FILE__, __LINE__, ERR_QUOTE(cond), ERR_QUOTE(ERROR(err))); \
-    _FORCE_HAS_FORMAT_STRING(__VA_ARGS__); \
-    RAWLOG(3, ": " __VA_ARGS__); \
-    RAWLOG(3, "\n"); \
-    return ERROR(err); \
-  }
+#define RETURN_ERROR_IF(cond, err, ...)                                        \
+    do {                                                                       \
+        if (cond) {                                                            \
+            RAWLOG(3, "%s:%d: ERROR!: check %s failed, returning %s",          \
+                  __FILE__, __LINE__, ERR_QUOTE(cond), ERR_QUOTE(ERROR(err))); \
+            _FORCE_HAS_FORMAT_STRING(__VA_ARGS__);                             \
+            RAWLOG(3, ": " __VA_ARGS__);                                       \
+            RAWLOG(3, "\n");                                                   \
+            return ERROR(err);                                                 \
+        }                                                                      \
+    } while (0)
 
 /**
  * Unconditionally return the specified error.
  *
  * In debug modes, prints additional information.
  */
-#define RETURN_ERROR(err, ...) \
-  do { \
-    RAWLOG(3, "%s:%d: ERROR!: unconditional check failed, returning %s", \
-           __FILE__, __LINE__, ERR_QUOTE(ERROR(err))); \
-    _FORCE_HAS_FORMAT_STRING(__VA_ARGS__); \
-    RAWLOG(3, ": " __VA_ARGS__); \
-    RAWLOG(3, "\n"); \
-    return ERROR(err); \
-  } while(0);
+#define RETURN_ERROR(err, ...)                                               \
+    do {                                                                     \
+        RAWLOG(3, "%s:%d: ERROR!: unconditional check failed, returning %s", \
+              __FILE__, __LINE__, ERR_QUOTE(ERROR(err)));                    \
+        _FORCE_HAS_FORMAT_STRING(__VA_ARGS__);                               \
+        RAWLOG(3, ": " __VA_ARGS__);                                         \
+        RAWLOG(3, "\n");                                                     \
+        return ERROR(err);                                                   \
+    } while(0)
 
 /**
  * If the provided expression evaluates to an error code, returns that error code.
  *
  * In debug modes, prints additional information.
  */
-#define FORWARD_IF_ERROR(err, ...) \
-  do { \
-    size_t const err_code = (err); \
-    if (ERR_isError(err_code)) { \
-      RAWLOG(3, "%s:%d: ERROR!: forwarding error in %s: %s", \
-             __FILE__, __LINE__, ERR_QUOTE(err), ERR_getErrorName(err_code)); \
-      _FORCE_HAS_FORMAT_STRING(__VA_ARGS__); \
-      RAWLOG(3, ": " __VA_ARGS__); \
-      RAWLOG(3, "\n"); \
-      return err_code; \
-    } \
-  } while(0);
+#define FORWARD_IF_ERROR(err, ...)                                                 \
+    do {                                                                           \
+        size_t const err_code = (err);                                             \
+        if (ERR_isError(err_code)) {                                               \
+            RAWLOG(3, "%s:%d: ERROR!: forwarding error in %s: %s",                 \
+                  __FILE__, __LINE__, ERR_QUOTE(err), ERR_getErrorName(err_code)); \
+            _FORCE_HAS_FORMAT_STRING(__VA_ARGS__);                                 \
+            RAWLOG(3, ": " __VA_ARGS__);                                           \
+            RAWLOG(3, "\n");                                                       \
+            return err_code;                                                       \
+        }                                                                          \
+    } while(0)
 
 #if defined (__cplusplus)
 }

lib/common/fse.h

@@ -229,6 +229,7 @@ If there is an error, the function will return an error code, which can be teste
 
 #endif  /* FSE_H */
 
+
 #if defined(FSE_STATIC_LINKING_ONLY) && !defined(FSE_H_FSE_STATIC_LINKING_ONLY)
 #define FSE_H_FSE_STATIC_LINKING_ONLY
 
@@ -464,13 +465,13 @@ MEM_STATIC void FSE_encodeSymbol(BIT_CStream_t* bitC, FSE_CState_t* statePtr, un
     FSE_symbolCompressionTransform const symbolTT = ((const FSE_symbolCompressionTransform*)(statePtr->symbolTT))[symbol];
     const U16* const stateTable = (const U16*)(statePtr->stateTable);
     U32 const nbBitsOut  = (U32)((statePtr->value + symbolTT.deltaNbBits) >> 16);
-    BIT_addBits(bitC, statePtr->value, nbBitsOut);
+    BIT_addBits(bitC,  (size_t)statePtr->value, nbBitsOut);
     statePtr->value = stateTable[ (statePtr->value >> nbBitsOut) + symbolTT.deltaFindState];
 }
 
 MEM_STATIC void FSE_flushCState(BIT_CStream_t* bitC, const FSE_CState_t* statePtr)
 {
-    BIT_addBits(bitC, statePtr->value, statePtr->stateLog);
+    BIT_addBits(bitC, (size_t)statePtr->value, statePtr->stateLog);
     BIT_flushBits(bitC);
 }
 

lib/common/fse_decompress.c

@@ -22,8 +22,7 @@
 #define FSE_STATIC_LINKING_ONLY
 #include "fse.h"
 #include "error_private.h"
-#define ZSTD_DEPS_NEED_MALLOC
-#include "zstd_deps.h"
+#include "zstd_deps.h"  /* ZSTD_memcpy */
 #include "bits.h"       /* ZSTD_highbit32 */
 
 
@@ -84,7 +83,7 @@ static size_t FSE_buildDTable_internal(FSE_DTable* dt, const short* normalizedCo
                     symbolNext[s] = 1;
                 } else {
                     if (normalizedCounter[s] >= largeLimit) DTableH.fastMode=0;
-                    symbolNext[s] = normalizedCounter[s];
+                    symbolNext[s] = (U16)normalizedCounter[s];
         }   }   }
         ZSTD_memcpy(dt, &DTableH, sizeof(DTableH));
     }
@@ -99,8 +98,7 @@ static size_t FSE_buildDTable_internal(FSE_DTable* dt, const short* normalizedCo
          * all symbols have counts <= 8. We ensure we have 8 bytes at the end of
          * our buffer to handle the over-write.
          */
-        {
-            U64 const add = 0x0101010101010101ull;
+        {   U64 const add = 0x0101010101010101ull;
             size_t pos = 0;
             U64 sv = 0;
             U32 s;
@@ -111,9 +109,8 @@ static size_t FSE_buildDTable_internal(FSE_DTable* dt, const short* normalizedCo
                 for (i = 8; i < n; i += 8) {
                     MEM_write64(spread + pos + i, sv);
                 }
-                pos += n;
-            }
-        }
+                pos += (size_t)n;
+        }   }
         /* Now we spread those positions across the table.
          * The benefit of doing it in two stages is that we avoid the
          * variable size inner loop, which caused lots of branch misses.
@@ -232,12 +229,12 @@ FORCE_INLINE_TEMPLATE size_t FSE_decompress_usingDTable_generic(
             break;
     }   }
 
-    return op-ostart;
+    assert(op >= ostart);
+    return (size_t)(op-ostart);
 }
 
 typedef struct {
     short ncount[FSE_MAX_SYMBOL_VALUE + 1];
-    FSE_DTable dtable[1]; /* Dynamically sized */
 } FSE_DecompressWksp;
 
 
@@ -252,13 +249,18 @@ FORCE_INLINE_TEMPLATE size_t FSE_decompress_wksp_body(
     unsigned tableLog;
     unsigned maxSymbolValue = FSE_MAX_SYMBOL_VALUE;
     FSE_DecompressWksp* const wksp = (FSE_DecompressWksp*)workSpace;
+    size_t const dtablePos = sizeof(FSE_DecompressWksp) / sizeof(FSE_DTable);
+    FSE_DTable* const dtable = (FSE_DTable*)workSpace + dtablePos;
 
-    DEBUG_STATIC_ASSERT((FSE_MAX_SYMBOL_VALUE + 1) % 2 == 0);
+    FSE_STATIC_ASSERT((FSE_MAX_SYMBOL_VALUE + 1) % 2 == 0);
     if (wkspSize < sizeof(*wksp)) return ERROR(GENERIC);
 
+    /* correct offset to dtable depends on this property */
+    FSE_STATIC_ASSERT(sizeof(FSE_DecompressWksp) % sizeof(FSE_DTable) == 0);
+
     /* normal FSE decoding mode */
-    {
-        size_t const NCountLength = FSE_readNCount_bmi2(wksp->ncount, &maxSymbolValue, &tableLog, istart, cSrcSize, bmi2);
+    {   size_t const NCountLength =
+            FSE_readNCount_bmi2(wksp->ncount, &maxSymbolValue, &tableLog, istart, cSrcSize, bmi2);
         if (FSE_isError(NCountLength)) return NCountLength;
         if (tableLog > maxLog) return ERROR(tableLog_tooLarge);
         assert(NCountLength <= cSrcSize);
@@ -271,16 +273,16 @@ FORCE_INLINE_TEMPLATE size_t FSE_decompress_wksp_body(
     workSpace = (BYTE*)workSpace + sizeof(*wksp) + FSE_DTABLE_SIZE(tableLog);
     wkspSize -= sizeof(*wksp) + FSE_DTABLE_SIZE(tableLog);
 
-    CHECK_F( FSE_buildDTable_internal(wksp->dtable, wksp->ncount, maxSymbolValue, tableLog, workSpace, wkspSize) );
+    CHECK_F( FSE_buildDTable_internal(dtable, wksp->ncount, maxSymbolValue, tableLog, workSpace, wkspSize) );
 
     {
-        const void* ptr = wksp->dtable;
+        const void* ptr = dtable;
         const FSE_DTableHeader* DTableH = (const FSE_DTableHeader*)ptr;
         const U32 fastMode = DTableH->fastMode;
 
         /* select fast mode (static) */
-        if (fastMode) return FSE_decompress_usingDTable_generic(dst, dstCapacity, ip, cSrcSize, wksp->dtable, 1);
-        return FSE_decompress_usingDTable_generic(dst, dstCapacity, ip, cSrcSize, wksp->dtable, 0);
+        if (fastMode) return FSE_decompress_usingDTable_generic(dst, dstCapacity, ip, cSrcSize, dtable, 1);
+        return FSE_decompress_usingDTable_generic(dst, dstCapacity, ip, cSrcSize, dtable, 0);
     }
 }
 

lib/common/huf.h

@@ -197,9 +197,22 @@ size_t HUF_readCTable (HUF_CElt* CTable, unsigned* maxSymbolValuePtr, const void
 
 /** HUF_getNbBitsFromCTable() :
  *  Read nbBits from CTable symbolTable, for symbol `symbolValue` presumed <= HUF_SYMBOLVALUE_MAX
- *  Note 1 : is not inlined, as HUF_CElt definition is private */
+ *  Note 1 : If symbolValue > HUF_readCTableHeader(symbolTable).maxSymbolValue, returns 0
+ *  Note 2 : is not inlined, as HUF_CElt definition is private
+ */
 U32 HUF_getNbBitsFromCTable(const HUF_CElt* symbolTable, U32 symbolValue);
 
+typedef struct {
+    BYTE tableLog;
+    BYTE maxSymbolValue;
+    BYTE unused[sizeof(size_t) - 2];
+} HUF_CTableHeader;
+
+/** HUF_readCTableHeader() :
+ *  @returns The header from the CTable specifying the tableLog and the maxSymbolValue.
+ */
+HUF_CTableHeader HUF_readCTableHeader(HUF_CElt const* ctable);
+
 /*
  * HUF_decompress() does the following:
  * 1. select the decompression algorithm (X1, X2) based on pre-computed heuristics

lib/common/mem.h

@@ -31,15 +31,6 @@ extern "C" {
 #   include <stdlib.h>  /* _byteswap_ulong */
 #   include <intrin.h>  /* _byteswap_* */
 #endif
-#if defined(__GNUC__)
-#  define MEM_STATIC static __inline __attribute__((unused))
-#elif defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */)
-#  define MEM_STATIC static inline
-#elif defined(_MSC_VER)
-#  define MEM_STATIC static __inline
-#else
-#  define MEM_STATIC static  /* this version may generate warnings for unused static functions; disable the relevant warning */
-#endif
 
 /*-**************************************************************
 *  Basic Types

lib/common/pool.c

@@ -10,9 +10,9 @@
 
 
 /* ======   Dependencies   ======= */
+#include "../common/allocations.h"  /* ZSTD_customCalloc, ZSTD_customFree */
 #include "zstd_deps.h" /* size_t */
 #include "debug.h"     /* assert */
-#include "zstd_internal.h"  /* ZSTD_customCalloc, ZSTD_customFree */
 #include "pool.h"
 
 /* ======   Compiler specifics   ====== */
@@ -223,7 +223,7 @@ static int POOL_resize_internal(POOL_ctx* ctx, size_t numThreads)
     {   ZSTD_pthread_t* const threadPool = (ZSTD_pthread_t*)ZSTD_customCalloc(numThreads * sizeof(ZSTD_pthread_t), ctx->customMem);
         if (!threadPool) return 1;
         /* replace existing thread pool */
-        ZSTD_memcpy(threadPool, ctx->threads, ctx->threadCapacity * sizeof(*threadPool));
+        ZSTD_memcpy(threadPool, ctx->threads, ctx->threadCapacity * sizeof(ZSTD_pthread_t));
         ZSTD_customFree(ctx->threads, ctx->customMem);
         ctx->threads = threadPool;
         /* Initialize additional threads */

lib/common/pool.h

@@ -47,7 +47,7 @@ void POOL_joinJobs(POOL_ctx* ctx);
 /*! POOL_resize() :
  *  Expands or shrinks pool's number of threads.
  *  This is more efficient than releasing + creating a new context,
- *  since it tries to preserve and re-use existing threads.
+ *  since it tries to preserve and reuse existing threads.
  * `numThreads` must be at least 1.
  * @return : 0 when resize was successful,
  *           !0 (typically 1) if there is an error.

lib/common/portability_macros.h

@@ -68,6 +68,8 @@
 /* Mark the internal assembly functions as hidden  */
 #ifdef __ELF__
 # define ZSTD_HIDE_ASM_FUNCTION(func) .hidden func
+#elif defined(__APPLE__)
+# define ZSTD_HIDE_ASM_FUNCTION(func) .private_extern func
 #else
 # define ZSTD_HIDE_ASM_FUNCTION(func)
 #endif

lib/common/threading.c

@@ -47,7 +47,7 @@ static unsigned __stdcall worker(void *arg)
     void* (*start_routine)(void*);
     void* thread_arg;
 
-    /* Inialized thread_arg and start_routine and signal main thread that we don't need it
+    /* Initialized thread_arg and start_routine and signal main thread that we don't need it
      * to wait any longer.
      */
     {
@@ -73,10 +73,12 @@ int ZSTD_pthread_create(ZSTD_pthread_t* thread, const void* unused,
     ZSTD_thread_params_t thread_param;
     (void)unused;
 
+    if (thread==NULL) return -1;
+    *thread = NULL;
+
     thread_param.start_routine = start_routine;
     thread_param.arg = arg;
     thread_param.initialized = 0;
-    *thread = NULL;
 
     /* Setup thread initialization synchronization */
     if(ZSTD_pthread_cond_init(&thread_param.initialized_cond, NULL)) {
@@ -91,7 +93,7 @@ int ZSTD_pthread_create(ZSTD_pthread_t* thread, const void* unused,
 
     /* Spawn thread */
     *thread = (HANDLE)_beginthreadex(NULL, 0, worker, &thread_param, 0, NULL);
-    if (!thread) {
+    if (*thread==NULL) {
         ZSTD_pthread_mutex_destroy(&thread_param.initialized_mutex);
         ZSTD_pthread_cond_destroy(&thread_param.initialized_cond);
         return errno;
@@ -137,6 +139,7 @@ int ZSTD_pthread_join(ZSTD_pthread_t thread)
 
 int ZSTD_pthread_mutex_init(ZSTD_pthread_mutex_t* mutex, pthread_mutexattr_t const* attr)
 {
+    assert(mutex != NULL);
     *mutex = (pthread_mutex_t*)ZSTD_malloc(sizeof(pthread_mutex_t));
     if (!*mutex)
         return 1;
@@ -145,6 +148,7 @@ int ZSTD_pthread_mutex_init(ZSTD_pthread_mutex_t* mutex, pthread_mutexattr_t con
 
 int ZSTD_pthread_mutex_destroy(ZSTD_pthread_mutex_t* mutex)
 {
+    assert(mutex != NULL);
     if (!*mutex)
         return 0;
     {
@@ -156,6 +160,7 @@ int ZSTD_pthread_mutex_destroy(ZSTD_pthread_mutex_t* mutex)
 
 int ZSTD_pthread_cond_init(ZSTD_pthread_cond_t* cond, pthread_condattr_t const* attr)
 {
+    assert(cond != NULL);
     *cond = (pthread_cond_t*)ZSTD_malloc(sizeof(pthread_cond_t));
     if (!*cond)
         return 1;
@@ -164,6 +169,7 @@ int ZSTD_pthread_cond_init(ZSTD_pthread_cond_t* cond, pthread_condattr_t const*
 
 int ZSTD_pthread_cond_destroy(ZSTD_pthread_cond_t* cond)
 {
+    assert(cond != NULL);
     if (!*cond)
         return 0;
     {

lib/common/xxhash.c

@@ -1,24 +1,18 @@
 /*
- *  xxHash - Fast Hash algorithm
- *  Copyright (c) Meta Platforms, Inc. and affiliates.
- *
- *  You can contact the author at :
- *  - xxHash homepage: https://cyan4973.github.io/xxHash/
- *  - xxHash source repository : https://github.com/Cyan4973/xxHash
+ * xxHash - Extremely Fast Hash algorithm
+ * Copyright (c) Yann Collet - Meta Platforms, Inc
  *
  * This source code is licensed under both the BSD-style license (found in the
  * LICENSE file in the root directory of this source tree) and the GPLv2 (found
  * in the COPYING file in the root directory of this source tree).
  * You may select, at your option, one of the above-listed licenses.
-*/
-
-
+ */
 
 /*
  * xxhash.c instantiates functions defined in xxhash.h
  */
 
-#define XXH_STATIC_LINKING_ONLY   /* access advanced declarations */
-#define XXH_IMPLEMENTATION   /* access definitions */
+#define XXH_STATIC_LINKING_ONLY /* access advanced declarations */
+#define XXH_IMPLEMENTATION      /* access definitions */
 
 #include "xxhash.h"

lib/common/zstd_common.c

@@ -14,7 +14,6 @@
 *  Dependencies
 ***************************************/
 #define ZSTD_DEPS_NEED_MALLOC
-#include "zstd_deps.h"   /* ZSTD_malloc, ZSTD_calloc, ZSTD_free, ZSTD_memset */
 #include "error_private.h"
 #include "zstd_internal.h"
 
@@ -47,37 +46,3 @@ ZSTD_ErrorCode ZSTD_getErrorCode(size_t code) { return ERR_getErrorCode(code); }
 /*! ZSTD_getErrorString() :
  *  provides error code string from enum */
 const char* ZSTD_getErrorString(ZSTD_ErrorCode code) { return ERR_getErrorString(code); }
-
-
-
-/*=**************************************************************
-*  Custom allocator
-****************************************************************/
-void* ZSTD_customMalloc(size_t size, ZSTD_customMem customMem)
-{
-    if (customMem.customAlloc)
-        return customMem.customAlloc(customMem.opaque, size);
-    return ZSTD_malloc(size);
-}
-
-void* ZSTD_customCalloc(size_t size, ZSTD_customMem customMem)
-{
-    if (customMem.customAlloc) {
-        /* calloc implemented as malloc+memset;
-         * not as efficient as calloc, but next best guess for custom malloc */
-        void* const ptr = customMem.customAlloc(customMem.opaque, size);
-        ZSTD_memset(ptr, 0, size);
-        return ptr;
-    }
-    return ZSTD_calloc(1, size);
-}
-
-void ZSTD_customFree(void* ptr, ZSTD_customMem customMem)
-{
-    if (ptr!=NULL) {
-        if (customMem.customFree)
-            customMem.customFree(customMem.opaque, ptr);
-        else
-            ZSTD_free(ptr);
-    }
-}

lib/common/zstd_internal.h

@@ -178,7 +178,7 @@ static void ZSTD_copy8(void* dst, const void* src) {
     ZSTD_memcpy(dst, src, 8);
 #endif
 }
-#define COPY8(d,s) { ZSTD_copy8(d,s); d+=8; s+=8; }
+#define COPY8(d,s) do { ZSTD_copy8(d,s); d+=8; s+=8; } while (0)
 
 /* Need to use memmove here since the literal buffer can now be located within
    the dst buffer. In circumstances where the op "catches up" to where the
@@ -198,7 +198,7 @@ static void ZSTD_copy16(void* dst, const void* src) {
     ZSTD_memcpy(dst, copy16_buf, 16);
 #endif
 }
-#define COPY16(d,s) { ZSTD_copy16(d,s); d+=16; s+=16; }
+#define COPY16(d,s) do { ZSTD_copy16(d,s); d+=16; s+=16; } while (0)
 
 #define WILDCOPY_OVERLENGTH 32
 #define WILDCOPY_VECLEN 16
@@ -227,7 +227,7 @@ void ZSTD_wildcopy(void* dst, const void* src, ptrdiff_t length, ZSTD_overlap_e
     if (ovtype == ZSTD_overlap_src_before_dst && diff < WILDCOPY_VECLEN) {
         /* Handle short offset copies. */
         do {
-            COPY8(op, ip)
+            COPY8(op, ip);
         } while (op < oend);
     } else {
         assert(diff >= WILDCOPY_VECLEN || diff <= -WILDCOPY_VECLEN);
@@ -350,11 +350,6 @@ typedef struct {
 const seqStore_t* ZSTD_getSeqStore(const ZSTD_CCtx* ctx);   /* compress & dictBuilder */
 int ZSTD_seqToCodes(const seqStore_t* seqStorePtr);   /* compress, dictBuilder, decodeCorpus (shouldn't get its definition from here) */
 
-/* custom memory allocation functions */
-void* ZSTD_customMalloc(size_t size, ZSTD_customMem customMem);
-void* ZSTD_customCalloc(size_t size, ZSTD_customMem customMem);
-void ZSTD_customFree(void* ptr, ZSTD_customMem customMem);
-
 
 /* ZSTD_invalidateRepCodes() :
  * ensures next compression will not use repcodes from previous block.
@@ -371,13 +366,13 @@ typedef struct {
 
 /*! ZSTD_getcBlockSize() :
  *  Provides the size of compressed block from block header `src` */
-/* Used by: decompress, fullbench (does not get its definition from here) */
+/*  Used by: decompress, fullbench */
 size_t ZSTD_getcBlockSize(const void* src, size_t srcSize,
                           blockProperties_t* bpPtr);
 
 /*! ZSTD_decodeSeqHeaders() :
  *  decode sequence header from src */
-/* Used by: decompress, fullbench (does not get its definition from here) */
+/*  Used by: zstd_decompress_block, fullbench */
 size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx* dctx, int* nbSeqPtr,
                        const void* src, size_t srcSize);
 

lib/compress/fse_compress.c

@@ -25,7 +25,7 @@
 #include "../common/error_private.h"
 #define ZSTD_DEPS_NEED_MALLOC
 #define ZSTD_DEPS_NEED_MATH64
-#include "../common/zstd_deps.h"  /* ZSTD_malloc, ZSTD_free, ZSTD_memcpy, ZSTD_memset */
+#include "../common/zstd_deps.h"  /* ZSTD_memset */
 #include "../common/bits.h" /* ZSTD_highbit32 */
 
 
@@ -225,8 +225,8 @@ size_t FSE_NCountWriteBound(unsigned maxSymbolValue, unsigned tableLog)
     size_t const maxHeaderSize = (((maxSymbolValue+1) * tableLog
                                    + 4 /* bitCount initialized at 4 */
                                    + 2 /* first two symbols may use one additional bit each */) / 8)
-                                    + 1 /* round up to whole nb bytes */
-                                    + 2 /* additional two bytes for bitstream flush */;
+                                   + 1 /* round up to whole nb bytes */
+                                   + 2 /* additional two bytes for bitstream flush */;
     return maxSymbolValue ? maxHeaderSize : FSE_NCOUNTBOUND;  /* maxSymbolValue==0 ? use default */
 }
 
@@ -255,7 +255,7 @@ FSE_writeNCount_generic (void* header, size_t headerBufferSize,
     /* Init */
     remaining = tableSize+1;   /* +1 for extra accuracy */
     threshold = tableSize;
-    nbBits = tableLog+1;
+    nbBits = (int)tableLog+1;
 
     while ((symbol < alphabetSize) && (remaining>1)) {  /* stops at 1 */
         if (previousIs0) {
@@ -274,7 +274,7 @@ FSE_writeNCount_generic (void* header, size_t headerBufferSize,
             }
             while (symbol >= start+3) {
                 start+=3;
-                bitStream += 3 << bitCount;
+                bitStream += 3U << bitCount;
                 bitCount += 2;
             }
             bitStream += (symbol-start) << bitCount;
@@ -294,7 +294,7 @@ FSE_writeNCount_generic (void* header, size_t headerBufferSize,
             count++;   /* +1 for extra accuracy */
             if (count>=threshold)
                 count += max;   /* [0..max[ [max..threshold[ (...) [threshold+max 2*threshold[ */
-            bitStream += count << bitCount;
+            bitStream += (U32)count << bitCount;
             bitCount  += nbBits;
             bitCount  -= (count<max);
             previousIs0  = (count==1);
@@ -322,7 +322,8 @@ FSE_writeNCount_generic (void* header, size_t headerBufferSize,
     out[1] = (BYTE)(bitStream>>8);
     out+= (bitCount+7) /8;
 
-    return (out-ostart);
+    assert(out >= ostart);
+    return (size_t)(out-ostart);
 }
 
 

lib/compress/huf_compress.c

@@ -220,6 +220,25 @@ static void HUF_setValue(HUF_CElt* elt, size_t value)
     }
 }
 
+HUF_CTableHeader HUF_readCTableHeader(HUF_CElt const* ctable)
+{
+    HUF_CTableHeader header;
+    ZSTD_memcpy(&header, ctable, sizeof(header));
+    return header;
+}
+
+static void HUF_writeCTableHeader(HUF_CElt* ctable, U32 tableLog, U32 maxSymbolValue)
+{
+    HUF_CTableHeader header;
+    HUF_STATIC_ASSERT(sizeof(ctable[0]) == sizeof(header));
+    ZSTD_memset(&header, 0, sizeof(header));
+    assert(tableLog < 256);
+    header.tableLog = (BYTE)tableLog;
+    assert(maxSymbolValue < 256);
+    header.maxSymbolValue = (BYTE)maxSymbolValue;
+    ZSTD_memcpy(ctable, &header, sizeof(header));
+}
+
 typedef struct {
     HUF_CompressWeightsWksp wksp;
     BYTE bitsToWeight[HUF_TABLELOG_MAX + 1];   /* precomputed conversion table */
@@ -237,6 +256,9 @@ size_t HUF_writeCTable_wksp(void* dst, size_t maxDstSize,
 
     HUF_STATIC_ASSERT(HUF_CTABLE_WORKSPACE_SIZE >= sizeof(HUF_WriteCTableWksp));
 
+    assert(HUF_readCTableHeader(CTable).maxSymbolValue == maxSymbolValue);
+    assert(HUF_readCTableHeader(CTable).tableLog == huffLog);
+
     /* check conditions */
     if (workspaceSize < sizeof(HUF_WriteCTableWksp)) return ERROR(GENERIC);
     if (maxSymbolValue > HUF_SYMBOLVALUE_MAX) return ERROR(maxSymbolValue_tooLarge);
@@ -283,7 +305,9 @@ size_t HUF_readCTable (HUF_CElt* CTable, unsigned* maxSymbolValuePtr, const void
     if (tableLog > HUF_TABLELOG_MAX) return ERROR(tableLog_tooLarge);
     if (nbSymbols > *maxSymbolValuePtr+1) return ERROR(maxSymbolValue_tooSmall);
 
-    CTable[0] = tableLog;
+    *maxSymbolValuePtr = nbSymbols - 1;
+
+    HUF_writeCTableHeader(CTable, tableLog, *maxSymbolValuePtr);
 
     /* Prepare base value per rank */
     {   U32 n, nextRankStart = 0;
@@ -315,7 +339,6 @@ size_t HUF_readCTable (HUF_CElt* CTable, unsigned* maxSymbolValuePtr, const void
         { U32 n; for (n=0; n<nbSymbols; n++) HUF_setValue(ct + n, valPerRank[HUF_getNbBits(ct[n])]++); }
     }
 
-    *maxSymbolValuePtr = nbSymbols - 1;
     return readSize;
 }
 
@@ -323,6 +346,8 @@ U32 HUF_getNbBitsFromCTable(HUF_CElt const* CTable, U32 symbolValue)
 {
     const HUF_CElt* const ct = CTable + 1;
     assert(symbolValue <= HUF_SYMBOLVALUE_MAX);
+    if (symbolValue > HUF_readCTableHeader(CTable).maxSymbolValue)
+        return 0;
     return (U32)HUF_getNbBits(ct[symbolValue]);
 }
 
@@ -723,7 +748,8 @@ static void HUF_buildCTableFromTree(HUF_CElt* CTable, nodeElt const* huffNode, i
         HUF_setNbBits(ct + huffNode[n].byte, huffNode[n].nbBits);   /* push nbBits per symbol, symbol order */
     for (n=0; n<alphabetSize; n++)
         HUF_setValue(ct + n, valPerRank[HUF_getNbBits(ct[n])]++);   /* assign value within rank, symbol order */
-    CTable[0] = maxNbBits;
+
+    HUF_writeCTableHeader(CTable, maxNbBits, maxSymbolValue);
 }
 
 size_t
@@ -776,13 +802,20 @@ size_t HUF_estimateCompressedSize(const HUF_CElt* CTable, const unsigned* count,
 }
 
 int HUF_validateCTable(const HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue) {
-  HUF_CElt const* ct = CTable + 1;
-  int bad = 0;
-  int s;
-  for (s = 0; s <= (int)maxSymbolValue; ++s) {
-    bad |= (count[s] != 0) & (HUF_getNbBits(ct[s]) == 0);
-  }
-  return !bad;
+    HUF_CTableHeader header = HUF_readCTableHeader(CTable);
+    HUF_CElt const* ct = CTable + 1;
+    int bad = 0;
+    int s;
+
+    assert(header.tableLog <= HUF_TABLELOG_ABSOLUTEMAX);
+
+    if (header.maxSymbolValue < maxSymbolValue)
+        return 0;
+
+    for (s = 0; s <= (int)maxSymbolValue; ++s) {
+        bad |= (count[s] != 0) & (HUF_getNbBits(ct[s]) == 0);
+    }
+    return !bad;
 }
 
 size_t HUF_compressBound(size_t size) { return HUF_COMPRESSBOUND(size); }
@@ -1024,17 +1057,17 @@ HUF_compress1X_usingCTable_internal_body(void* dst, size_t dstSize,
                                    const void* src, size_t srcSize,
                                    const HUF_CElt* CTable)
 {
-    U32 const tableLog = (U32)CTable[0];
+    U32 const tableLog = HUF_readCTableHeader(CTable).tableLog;
     HUF_CElt const* ct = CTable + 1;
     const BYTE* ip = (const BYTE*) src;
     BYTE* const ostart = (BYTE*)dst;
     BYTE* const oend = ostart + dstSize;
-    BYTE* op = ostart;
     HUF_CStream_t bitC;
 
     /* init */
     if (dstSize < 8) return 0;   /* not enough space to compress */
-    { size_t const initErr = HUF_initCStream(&bitC, op, (size_t)(oend-op));
+    { BYTE* op = ostart;
+      size_t const initErr = HUF_initCStream(&bitC, op, (size_t)(oend-op));
       if (HUF_isError(initErr)) return 0; }
 
     if (dstSize < HUF_tightCompressBound(srcSize, (size_t)tableLog) || tableLog > 11)
@@ -1255,7 +1288,7 @@ unsigned HUF_optimalTableLog(
 
     {   BYTE* dst = (BYTE*)workSpace + sizeof(HUF_WriteCTableWksp);
         size_t dstSize = wkspSize - sizeof(HUF_WriteCTableWksp);
-        size_t maxBits, hSize, newSize;
+        size_t hSize, newSize;
         const unsigned symbolCardinality = HUF_cardinality(count, maxSymbolValue);
         const unsigned minTableLog = HUF_minTableLog(symbolCardinality);
         size_t optSize = ((size_t) ~0) - 1;
@@ -1266,12 +1299,14 @@ unsigned HUF_optimalTableLog(
         /* Search until size increases */
         for (optLogGuess = minTableLog; optLogGuess <= maxTableLog; optLogGuess++) {
             DEBUGLOG(7, "checking for huffLog=%u", optLogGuess);
-            maxBits = HUF_buildCTable_wksp(table, count, maxSymbolValue, optLogGuess, workSpace, wkspSize);
-            if (ERR_isError(maxBits)) continue;
 
-            if (maxBits < optLogGuess && optLogGuess > minTableLog) break;
+            {   size_t maxBits = HUF_buildCTable_wksp(table, count, maxSymbolValue, optLogGuess, workSpace, wkspSize);
+                if (ERR_isError(maxBits)) continue;
 
-            hSize = HUF_writeCTable_wksp(dst, dstSize, table, maxSymbolValue, (U32)maxBits, workSpace, wkspSize);
+                if (maxBits < optLogGuess && optLogGuess > minTableLog) break;
+
+                hSize = HUF_writeCTable_wksp(dst, dstSize, table, maxSymbolValue, (U32)maxBits, workSpace, wkspSize);
+            }
 
             if (ERR_isError(hSize)) continue;
 
@@ -1372,12 +1407,6 @@ HUF_compress_internal (void* dst, size_t dstSize,
         huffLog = (U32)maxBits;
         DEBUGLOG(6, "bit distribution completed (%zu symbols)", showCTableBits(table->CTable + 1, maxSymbolValue+1));
     }
-    /* Zero unused symbols in CTable, so we can check it for validity */
-    {
-        size_t const ctableSize = HUF_CTABLE_SIZE_ST(maxSymbolValue);
-        size_t const unusedSize = sizeof(table->CTable) - ctableSize * sizeof(HUF_CElt);
-        ZSTD_memset(table->CTable + ctableSize, 0, unusedSize);
-    }
 
     /* Write table description header */
     {   CHECK_V_F(hSize, HUF_writeCTable_wksp(op, dstSize, table->CTable, maxSymbolValue, huffLog,
@@ -1420,7 +1449,7 @@ size_t HUF_compress1X_repeat (void* dst, size_t dstSize,
 /* HUF_compress4X_repeat():
  * compress input using 4 streams.
  * consider skipping quickly
- * re-use an existing huffman compression table */
+ * reuse an existing huffman compression table */
 size_t HUF_compress4X_repeat (void* dst, size_t dstSize,
                       const void* src, size_t srcSize,
                       unsigned maxSymbolValue, unsigned huffLog,

lib/compress/zstd_compress_internal.h

@@ -39,7 +39,7 @@ extern "C" {
                                        It's not a big deal though : candidate will just be sorted again.
                                        Additionally, candidate position 1 will be lost.
                                        But candidate 1 cannot hide a large tree of candidates, so it's a minimal loss.
-                                       The benefit is that ZSTD_DUBT_UNSORTED_MARK cannot be mishandled after table re-use with a different strategy.
+                                       The benefit is that ZSTD_DUBT_UNSORTED_MARK cannot be mishandled after table reuse with a different strategy.
                                        This constant is required by ZSTD_compressBlock_btlazy2() and ZSTD_reduceTable_internal() */
 
 
@@ -159,23 +159,24 @@ typedef struct {
 UNUSED_ATTR static const rawSeqStore_t kNullRawSeqStore = {NULL, 0, 0, 0, 0};
 
 typedef struct {
-    int price;
-    U32 off;
-    U32 mlen;
-    U32 litlen;
-    U32 rep[ZSTD_REP_NUM];
+    int price;  /* price from beginning of segment to this position */
+    U32 off;    /* offset of previous match */
+    U32 mlen;   /* length of previous match */
+    U32 litlen; /* nb of literals since previous match */
+    U32 rep[ZSTD_REP_NUM];  /* offset history after previous match */
 } ZSTD_optimal_t;
 
 typedef enum { zop_dynamic=0, zop_predef } ZSTD_OptPrice_e;
 
+#define ZSTD_OPT_SIZE (ZSTD_OPT_NUM+3)
 typedef struct {
     /* All tables are allocated inside cctx->workspace by ZSTD_resetCCtx_internal() */
     unsigned* litFreq;           /* table of literals statistics, of size 256 */
     unsigned* litLengthFreq;     /* table of litLength statistics, of size (MaxLL+1) */
     unsigned* matchLengthFreq;   /* table of matchLength statistics, of size (MaxML+1) */
     unsigned* offCodeFreq;       /* table of offCode statistics, of size (MaxOff+1) */
-    ZSTD_match_t* matchTable;    /* list of found matches, of size ZSTD_OPT_NUM+1 */
-    ZSTD_optimal_t* priceTable;  /* All positions tracked by optimal parser, of size ZSTD_OPT_NUM+1 */
+    ZSTD_match_t* matchTable;    /* list of found matches, of size ZSTD_OPT_SIZE */
+    ZSTD_optimal_t* priceTable;  /* All positions tracked by optimal parser, of size ZSTD_OPT_SIZE */
 
     U32  litSum;                 /* nb of literals */
     U32  litLengthSum;           /* nb of litLength codes */
@@ -226,8 +227,10 @@ struct ZSTD_matchState_t {
     U32 hashLog3;           /* dispatch table for matches of len==3 : larger == faster, more memory */
 
     U32 rowHashLog;                          /* For row-based matchfinder: Hashlog based on nb of rows in the hashTable.*/
-    U16* tagTable;                           /* For row-based matchFinder: A row-based table containing the hashes and head index. */
+    BYTE* tagTable;                          /* For row-based matchFinder: A row-based table containing the hashes and head index. */
     U32 hashCache[ZSTD_ROW_HASH_CACHE_SIZE]; /* For row-based matchFinder: a cache of hashes to improve speed */
+    U64 hashSalt;                            /* For row-based matchFinder: salts the hash for reuse of tag table */
+    U32 hashSaltEntropy;                     /* For row-based matchFinder: collects entropy for salt generation */
 
     U32* hashTable;
     U32* hashTable3;
@@ -247,6 +250,13 @@ struct ZSTD_matchState_t {
      * This behavior is controlled from the cctx ms.
      * This parameter has no effect in the cdict ms. */
     int prefetchCDictTables;
+
+    /* When == 0, lazy match finders insert every position.
+     * When != 0, lazy match finders only insert positions they search.
+     * This allows them to skip much faster over incompressible data,
+     * at a small cost to compression ratio.
+     */
+    int lazySkipping;
 };
 
 typedef struct {
@@ -351,10 +361,11 @@ struct ZSTD_CCtx_params_s {
      * if the external matchfinder returns an error code. */
     int enableMatchFinderFallback;
 
-    /* Indicates whether an external matchfinder has been referenced.
-     * Users can't set this externally.
-     * It is set internally in ZSTD_registerSequenceProducer(). */
-    int useSequenceProducer;
+    /* Parameters for the external sequence producer API.
+     * Users set these parameters through ZSTD_registerSequenceProducer().
+     * It is not possible to set these parameters individually through the public API. */
+    void* extSeqProdState;
+    ZSTD_sequenceProducer_F extSeqProdFunc;
 
     /* Adjust the max block size*/
     size_t maxBlockSize;
@@ -392,14 +403,6 @@ typedef struct {
     ZSTD_entropyCTablesMetadata_t entropyMetadata;
 } ZSTD_blockSplitCtx;
 
-/* Context for block-level external matchfinder API */
-typedef struct {
-  void* mState;
-  ZSTD_sequenceProducer_F* mFinder;
-  ZSTD_Sequence* seqBuffer;
-  size_t seqBufferCapacity;
-} ZSTD_externalMatchCtx;
-
 struct ZSTD_CCtx_s {
     ZSTD_compressionStage_e stage;
     int cParamsChanged;                  /* == 1 if cParams(except wlog) or compression level are changed in requestedParams. Triggers transmission of new params to ZSTDMT (if available) then reset to 0. */
@@ -470,8 +473,9 @@ struct ZSTD_CCtx_s {
     /* Workspace for block splitter */
     ZSTD_blockSplitCtx blockSplitCtx;
 
-    /* Workspace for external matchfinder */
-    ZSTD_externalMatchCtx externalMatchCtx;
+    /* Buffer for output from external sequence producer */
+    ZSTD_Sequence* extSeqBuf;
+    size_t extSeqBufCapacity;
 };
 
 typedef enum { ZSTD_dtlm_fast, ZSTD_dtlm_full } ZSTD_dictTableLoadMethod_e;
@@ -787,28 +791,35 @@ ZSTD_count_2segments(const BYTE* ip, const BYTE* match,
  *  Hashes
  ***************************************/
 static const U32 prime3bytes = 506832829U;
-static U32    ZSTD_hash3(U32 u, U32 h) { assert(h <= 32); return ((u << (32-24)) * prime3bytes)  >> (32-h) ; }
-MEM_STATIC size_t ZSTD_hash3Ptr(const void* ptr, U32 h) { return ZSTD_hash3(MEM_readLE32(ptr), h); } /* only in zstd_opt.h */
+static U32    ZSTD_hash3(U32 u, U32 h, U32 s) { assert(h <= 32); return (((u << (32-24)) * prime3bytes) ^ s)  >> (32-h) ; }
+MEM_STATIC size_t ZSTD_hash3Ptr(const void* ptr, U32 h) { return ZSTD_hash3(MEM_readLE32(ptr), h, 0); } /* only in zstd_opt.h */
+MEM_STATIC size_t ZSTD_hash3PtrS(const void* ptr, U32 h, U32 s) { return ZSTD_hash3(MEM_readLE32(ptr), h, s); }
 
 static const U32 prime4bytes = 2654435761U;
-static U32    ZSTD_hash4(U32 u, U32 h) { assert(h <= 32); return (u * prime4bytes) >> (32-h) ; }
-static size_t ZSTD_hash4Ptr(const void* ptr, U32 h) { return ZSTD_hash4(MEM_readLE32(ptr), h); }
+static U32    ZSTD_hash4(U32 u, U32 h, U32 s) { assert(h <= 32); return ((u * prime4bytes) ^ s) >> (32-h) ; }
+static size_t ZSTD_hash4Ptr(const void* ptr, U32 h) { return ZSTD_hash4(MEM_readLE32(ptr), h, 0); }
+static size_t ZSTD_hash4PtrS(const void* ptr, U32 h, U32 s) { return ZSTD_hash4(MEM_readLE32(ptr), h, s); }
 
 static const U64 prime5bytes = 889523592379ULL;
-static size_t ZSTD_hash5(U64 u, U32 h) { assert(h <= 64); return (size_t)(((u  << (64-40)) * prime5bytes) >> (64-h)) ; }
-static size_t ZSTD_hash5Ptr(const void* p, U32 h) { return ZSTD_hash5(MEM_readLE64(p), h); }
+static size_t ZSTD_hash5(U64 u, U32 h, U64 s) { assert(h <= 64); return (size_t)((((u  << (64-40)) * prime5bytes) ^ s) >> (64-h)) ; }
+static size_t ZSTD_hash5Ptr(const void* p, U32 h) { return ZSTD_hash5(MEM_readLE64(p), h, 0); }
+static size_t ZSTD_hash5PtrS(const void* p, U32 h, U64 s) { return ZSTD_hash5(MEM_readLE64(p), h, s); }
 
 static const U64 prime6bytes = 227718039650203ULL;
-static size_t ZSTD_hash6(U64 u, U32 h) { assert(h <= 64); return (size_t)(((u  << (64-48)) * prime6bytes) >> (64-h)) ; }
-static size_t ZSTD_hash6Ptr(const void* p, U32 h) { return ZSTD_hash6(MEM_readLE64(p), h); }
+static size_t ZSTD_hash6(U64 u, U32 h, U64 s) { assert(h <= 64); return (size_t)((((u  << (64-48)) * prime6bytes) ^ s) >> (64-h)) ; }
+static size_t ZSTD_hash6Ptr(const void* p, U32 h) { return ZSTD_hash6(MEM_readLE64(p), h, 0); }
+static size_t ZSTD_hash6PtrS(const void* p, U32 h, U64 s) { return ZSTD_hash6(MEM_readLE64(p), h, s); }
 
 static const U64 prime7bytes = 58295818150454627ULL;
-static size_t ZSTD_hash7(U64 u, U32 h) { assert(h <= 64); return (size_t)(((u  << (64-56)) * prime7bytes) >> (64-h)) ; }
-static size_t ZSTD_hash7Ptr(const void* p, U32 h) { return ZSTD_hash7(MEM_readLE64(p), h); }
+static size_t ZSTD_hash7(U64 u, U32 h, U64 s) { assert(h <= 64); return (size_t)((((u  << (64-56)) * prime7bytes) ^ s) >> (64-h)) ; }
+static size_t ZSTD_hash7Ptr(const void* p, U32 h) { return ZSTD_hash7(MEM_readLE64(p), h, 0); }
+static size_t ZSTD_hash7PtrS(const void* p, U32 h, U64 s) { return ZSTD_hash7(MEM_readLE64(p), h, s); }
 
 static const U64 prime8bytes = 0xCF1BBCDCB7A56463ULL;
-static size_t ZSTD_hash8(U64 u, U32 h) { assert(h <= 64); return (size_t)(((u) * prime8bytes) >> (64-h)) ; }
-static size_t ZSTD_hash8Ptr(const void* p, U32 h) { return ZSTD_hash8(MEM_readLE64(p), h); }
+static size_t ZSTD_hash8(U64 u, U32 h, U64 s) { assert(h <= 64); return (size_t)((((u) * prime8bytes)  ^ s) >> (64-h)) ; }
+static size_t ZSTD_hash8Ptr(const void* p, U32 h) { return ZSTD_hash8(MEM_readLE64(p), h, 0); }
+static size_t ZSTD_hash8PtrS(const void* p, U32 h, U64 s) { return ZSTD_hash8(MEM_readLE64(p), h, s); }
+
 
 MEM_STATIC FORCE_INLINE_ATTR
 size_t ZSTD_hashPtr(const void* p, U32 hBits, U32 mls)
@@ -828,6 +839,24 @@ size_t ZSTD_hashPtr(const void* p, U32 hBits, U32 mls)
     }
 }
 
+MEM_STATIC FORCE_INLINE_ATTR
+size_t ZSTD_hashPtrSalted(const void* p, U32 hBits, U32 mls, const U64 hashSalt) {
+    /* Although some of these hashes do support hBits up to 64, some do not.
+     * To be on the safe side, always avoid hBits > 32. */
+    assert(hBits <= 32);
+
+    switch(mls)
+    {
+        default:
+        case 4: return ZSTD_hash4PtrS(p, hBits, (U32)hashSalt);
+        case 5: return ZSTD_hash5PtrS(p, hBits, hashSalt);
+        case 6: return ZSTD_hash6PtrS(p, hBits, hashSalt);
+        case 7: return ZSTD_hash7PtrS(p, hBits, hashSalt);
+        case 8: return ZSTD_hash8PtrS(p, hBits, hashSalt);
+    }
+}
+
+
 /** ZSTD_ipow() :
  * Return base^exponent.
  */
@@ -1019,7 +1048,9 @@ MEM_STATIC U32 ZSTD_window_needOverflowCorrection(ZSTD_window_t const window,
  * The least significant cycleLog bits of the indices must remain the same,
  * which may be 0. Every index up to maxDist in the past must be valid.
  */
-MEM_STATIC U32 ZSTD_window_correctOverflow(ZSTD_window_t* window, U32 cycleLog,
+MEM_STATIC
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+U32 ZSTD_window_correctOverflow(ZSTD_window_t* window, U32 cycleLog,
                                            U32 maxDist, void const* src)
 {
     /* preemptive overflow correction:
@@ -1212,7 +1243,9 @@ MEM_STATIC void ZSTD_window_init(ZSTD_window_t* window) {
  * forget about the extDict. Handles overlap of the prefix and extDict.
  * Returns non-zero if the segment is contiguous.
  */
-MEM_STATIC U32 ZSTD_window_update(ZSTD_window_t* window,
+MEM_STATIC
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+U32 ZSTD_window_update(ZSTD_window_t* window,
                                   void const* src, size_t srcSize,
                                   int forceNonContiguous)
 {
@@ -1433,11 +1466,10 @@ size_t ZSTD_writeLastEmptyBlock(void* dst, size_t dstCapacity);
  * This cannot be used when long range matching is enabled.
  * Zstd will use these sequences, and pass the literals to a secondary block
  * compressor.
- * @return : An error code on failure.
  * NOTE: seqs are not verified! Invalid sequences can cause out-of-bounds memory
  * access and data corruption.
  */
-size_t ZSTD_referenceExternalSequences(ZSTD_CCtx* cctx, rawSeq* seq, size_t nbSeq);
+void ZSTD_referenceExternalSequences(ZSTD_CCtx* cctx, rawSeq* seq, size_t nbSeq);
 
 /** ZSTD_cycleLog() :
  *  condition for correct operation : hashLog > 1 */
@@ -1475,4 +1507,28 @@ ZSTD_copySequencesToSeqStoreNoBlockDelim(ZSTD_CCtx* cctx, ZSTD_sequencePosition*
                                    const ZSTD_Sequence* const inSeqs, size_t inSeqsSize,
                                    const void* src, size_t blockSize, ZSTD_paramSwitch_e externalRepSearch);
 
+/* Returns 1 if an external sequence producer is registered, otherwise returns 0. */
+MEM_STATIC int ZSTD_hasExtSeqProd(const ZSTD_CCtx_params* params) {
+    return params->extSeqProdFunc != NULL;
+}
+
+/* ===============================================================
+ * Deprecated definitions that are still used internally to avoid
+ * deprecation warnings. These functions are exactly equivalent to
+ * their public variants, but avoid the deprecation warnings.
+ * =============================================================== */
+
+size_t ZSTD_compressBegin_usingCDict_deprecated(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict);
+
+size_t ZSTD_compressContinue_public(ZSTD_CCtx* cctx,
+                                    void* dst, size_t dstCapacity,
+                              const void* src, size_t srcSize);
+
+size_t ZSTD_compressEnd_public(ZSTD_CCtx* cctx,
+                               void* dst, size_t dstCapacity,
+                         const void* src, size_t srcSize);
+
+size_t ZSTD_compressBlock_deprecated(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+
+
 #endif /* ZSTD_COMPRESS_H */

lib/compress/zstd_compress_superblock.c

@@ -76,8 +76,8 @@ ZSTD_compressSubBlock_literal(const HUF_CElt* hufTable,
     }
 
     {   int const flags = bmi2 ? HUF_flags_bmi2 : 0;
-        const size_t cSize = singleStream ? HUF_compress1X_usingCTable(op, oend-op, literals, litSize, hufTable, flags)
-                                          : HUF_compress4X_usingCTable(op, oend-op, literals, litSize, hufTable, flags);
+        const size_t cSize = singleStream ? HUF_compress1X_usingCTable(op, (size_t)(oend-op), literals, litSize, hufTable, flags)
+                                          : HUF_compress4X_usingCTable(op, (size_t)(oend-op), literals, litSize, hufTable, flags);
         op += cSize;
         cLitSize += cSize;
         if (cSize == 0 || ERR_isError(cSize)) {
@@ -102,7 +102,7 @@ ZSTD_compressSubBlock_literal(const HUF_CElt* hufTable,
     switch(lhSize)
     {
     case 3: /* 2 - 2 - 10 - 10 */
-        {   U32 const lhc = hType + ((!singleStream) << 2) + ((U32)litSize<<4) + ((U32)cLitSize<<14);
+        {   U32 const lhc = hType + ((U32)(!singleStream) << 2) + ((U32)litSize<<4) + ((U32)cLitSize<<14);
             MEM_writeLE24(ostart, lhc);
             break;
         }
@@ -122,30 +122,30 @@ ZSTD_compressSubBlock_literal(const HUF_CElt* hufTable,
     }
     *entropyWritten = 1;
     DEBUGLOG(5, "Compressed literals: %u -> %u", (U32)litSize, (U32)(op-ostart));
-    return op-ostart;
+    return (size_t)(op-ostart);
 }
 
 static size_t
 ZSTD_seqDecompressedSize(seqStore_t const* seqStore,
-                   const seqDef* sequences, size_t nbSeq,
-                         size_t litSize, int lastSequence)
+                   const seqDef* sequences, size_t nbSeqs,
+                         size_t litSize, int lastSubBlock)
 {
-    const seqDef* const sstart = sequences;
-    const seqDef* const send = sequences + nbSeq;
-    const seqDef* sp = sstart;
     size_t matchLengthSum = 0;
     size_t litLengthSum = 0;
-    (void)(litLengthSum); /* suppress unused variable warning on some environments */
-    while (send-sp > 0) {
-        ZSTD_sequenceLength const seqLen = ZSTD_getSequenceLength(seqStore, sp);
+    size_t n;
+    for (n=0; n<nbSeqs; n++) {
+        const ZSTD_sequenceLength seqLen = ZSTD_getSequenceLength(seqStore, sequences+n);
         litLengthSum += seqLen.litLength;
         matchLengthSum += seqLen.matchLength;
-        sp++;
     }
-    assert(litLengthSum <= litSize);
-    if (!lastSequence) {
+    DEBUGLOG(5, "ZSTD_seqDecompressedSize: %u sequences from %p: %u literals + %u matchlength",
+                (unsigned)nbSeqs, (const void*)sequences,
+                (unsigned)litLengthSum, (unsigned)matchLengthSum);
+    if (!lastSubBlock)
         assert(litLengthSum == litSize);
-    }
+    else
+        assert(litLengthSum <= litSize);
+    (void)litLengthSum;
     return matchLengthSum + litSize;
 }
 
@@ -180,14 +180,14 @@ ZSTD_compressSubBlock_sequences(const ZSTD_fseCTables_t* fseTables,
     /* Sequences Header */
     RETURN_ERROR_IF((oend-op) < 3 /*max nbSeq Size*/ + 1 /*seqHead*/,
                     dstSize_tooSmall, "");
-    if (nbSeq < 0x7F)
+    if (nbSeq < 128)
         *op++ = (BYTE)nbSeq;
     else if (nbSeq < LONGNBSEQ)
         op[0] = (BYTE)((nbSeq>>8) + 0x80), op[1] = (BYTE)nbSeq, op+=2;
     else
         op[0]=0xFF, MEM_writeLE16(op+1, (U16)(nbSeq - LONGNBSEQ)), op+=3;
     if (nbSeq==0) {
-        return op - ostart;
+        return (size_t)(op - ostart);
     }
 
     /* seqHead : flags for FSE encoding type */
@@ -209,7 +209,7 @@ ZSTD_compressSubBlock_sequences(const ZSTD_fseCTables_t* fseTables,
     }
 
     {   size_t const bitstreamSize = ZSTD_encodeSequences(
-                                        op, oend - op,
+                                        op, (size_t)(oend - op),
                                         fseTables->matchlengthCTable, mlCode,
                                         fseTables->offcodeCTable, ofCode,
                                         fseTables->litlengthCTable, llCode,
@@ -253,7 +253,7 @@ ZSTD_compressSubBlock_sequences(const ZSTD_fseCTables_t* fseTables,
 #endif
 
     *entropyWritten = 1;
-    return op - ostart;
+    return (size_t)(op - ostart);
 }
 
 /** ZSTD_compressSubBlock() :
@@ -279,7 +279,8 @@ static size_t ZSTD_compressSubBlock(const ZSTD_entropyCTables_t* entropy,
                 litSize, nbSeq, writeLitEntropy, writeSeqEntropy, lastBlock);
     {   size_t cLitSize = ZSTD_compressSubBlock_literal((const HUF_CElt*)entropy->huf.CTable,
                                                         &entropyMetadata->hufMetadata, literals, litSize,
-                                                        op, oend-op, bmi2, writeLitEntropy, litEntropyWritten);
+                                                        op, (size_t)(oend-op),
+                                                        bmi2, writeLitEntropy, litEntropyWritten);
         FORWARD_IF_ERROR(cLitSize, "ZSTD_compressSubBlock_literal failed");
         if (cLitSize == 0) return 0;
         op += cLitSize;
@@ -289,18 +290,18 @@ static size_t ZSTD_compressSubBlock(const ZSTD_entropyCTables_t* entropy,
                                                   sequences, nbSeq,
                                                   llCode, mlCode, ofCode,
                                                   cctxParams,
-                                                  op, oend-op,
+                                                  op, (size_t)(oend-op),
                                                   bmi2, writeSeqEntropy, seqEntropyWritten);
         FORWARD_IF_ERROR(cSeqSize, "ZSTD_compressSubBlock_sequences failed");
         if (cSeqSize == 0) return 0;
         op += cSeqSize;
     }
     /* Write block header */
-    {   size_t cSize = (op-ostart)-ZSTD_blockHeaderSize;
+    {   size_t cSize = (size_t)(op-ostart) - ZSTD_blockHeaderSize;
         U32 const cBlockHeader24 = lastBlock + (((U32)bt_compressed)<<1) + (U32)(cSize << 3);
         MEM_writeLE24(ostart, cBlockHeader24);
     }
-    return op-ostart;
+    return (size_t)(op-ostart);
 }
 
 static size_t ZSTD_estimateSubBlockSize_literal(const BYTE* literals, size_t litSize,
@@ -389,7 +390,11 @@ static size_t ZSTD_estimateSubBlockSize_sequences(const BYTE* ofCodeTable,
     return cSeqSizeEstimate + sequencesSectionHeaderSize;
 }
 
-static size_t ZSTD_estimateSubBlockSize(const BYTE* literals, size_t litSize,
+typedef struct {
+    size_t estLitSize;
+    size_t estBlockSize;
+} EstimatedBlockSize;
+static EstimatedBlockSize ZSTD_estimateSubBlockSize(const BYTE* literals, size_t litSize,
                                         const BYTE* ofCodeTable,
                                         const BYTE* llCodeTable,
                                         const BYTE* mlCodeTable,
@@ -397,15 +402,17 @@ static size_t ZSTD_estimateSubBlockSize(const BYTE* literals, size_t litSize,
                                         const ZSTD_entropyCTables_t* entropy,
                                         const ZSTD_entropyCTablesMetadata_t* entropyMetadata,
                                         void* workspace, size_t wkspSize,
-                                        int writeLitEntropy, int writeSeqEntropy) {
-    size_t cSizeEstimate = 0;
-    cSizeEstimate += ZSTD_estimateSubBlockSize_literal(literals, litSize,
-                                                         &entropy->huf, &entropyMetadata->hufMetadata,
-                                                         workspace, wkspSize, writeLitEntropy);
-    cSizeEstimate += ZSTD_estimateSubBlockSize_sequences(ofCodeTable, llCodeTable, mlCodeTable,
+                                        int writeLitEntropy, int writeSeqEntropy)
+{
+    EstimatedBlockSize ebs;
+    ebs.estLitSize = ZSTD_estimateSubBlockSize_literal(literals, litSize,
+                                                        &entropy->huf, &entropyMetadata->hufMetadata,
+                                                        workspace, wkspSize, writeLitEntropy);
+    ebs.estBlockSize = ZSTD_estimateSubBlockSize_sequences(ofCodeTable, llCodeTable, mlCodeTable,
                                                          nbSeq, &entropy->fse, &entropyMetadata->fseMetadata,
                                                          workspace, wkspSize, writeSeqEntropy);
-    return cSizeEstimate + ZSTD_blockHeaderSize;
+    ebs.estBlockSize += ebs.estLitSize + ZSTD_blockHeaderSize;
+    return ebs;
 }
 
 static int ZSTD_needSequenceEntropyTables(ZSTD_fseCTablesMetadata_t const* fseMetadata)
@@ -419,13 +426,56 @@ static int ZSTD_needSequenceEntropyTables(ZSTD_fseCTablesMetadata_t const* fseMe
     return 0;
 }
 
+static size_t countLiterals(seqStore_t const* seqStore, const seqDef* sp, size_t seqCount)
+{
+    size_t n, total = 0;
+    assert(sp != NULL);
+    for (n=0; n<seqCount; n++) {
+        total += ZSTD_getSequenceLength(seqStore, sp+n).litLength;
+    }
+    DEBUGLOG(6, "countLiterals for %zu sequences from %p => %zu bytes", seqCount, (const void*)sp, total);
+    return total;
+}
+
+#define BYTESCALE 256
+
+static size_t sizeBlockSequences(const seqDef* sp, size_t nbSeqs,
+                size_t targetBudget, size_t avgLitCost, size_t avgSeqCost,
+                int firstSubBlock)
+{
+    size_t n, budget = 0, inSize=0;
+    /* entropy headers */
+    size_t const headerSize = (size_t)firstSubBlock * 120 * BYTESCALE; /* generous estimate */
+    assert(firstSubBlock==0 || firstSubBlock==1);
+    budget += headerSize;
+
+    /* first sequence => at least one sequence*/
+    budget += sp[0].litLength * avgLitCost + avgSeqCost;
+    if (budget > targetBudget) return 1;
+    inSize = sp[0].litLength + (sp[0].mlBase+MINMATCH);
+
+    /* loop over sequences */
+    for (n=1; n<nbSeqs; n++) {
+        size_t currentCost = sp[n].litLength * avgLitCost + avgSeqCost;
+        budget += currentCost;
+        inSize += sp[n].litLength + (sp[n].mlBase+MINMATCH);
+        /* stop when sub-block budget is reached */
+        if ( (budget > targetBudget)
+            /* though continue to expand until the sub-block is deemed compressible */
+          && (budget < inSize * BYTESCALE) )
+            break;
+    }
+
+    return n;
+}
+
 /** ZSTD_compressSubBlock_multi() :
  *  Breaks super-block into multiple sub-blocks and compresses them.
- *  Entropy will be written to the first block.
- *  The following blocks will use repeat mode to compress.
- *  All sub-blocks are compressed blocks (no raw or rle blocks).
- *  @return : compressed size of the super block (which is multiple ZSTD blocks)
- *            Or 0 if it failed to compress. */
+ *  Entropy will be written into the first block.
+ *  The following blocks use repeat_mode to compress.
+ *  Sub-blocks are all compressed, except the last one when beneficial.
+ *  @return : compressed size of the super block (which features multiple ZSTD blocks)
+ *            or 0 if it failed to compress. */
 static size_t ZSTD_compressSubBlock_multi(const seqStore_t* seqStorePtr,
                             const ZSTD_compressedBlockState_t* prevCBlock,
                             ZSTD_compressedBlockState_t* nextCBlock,
@@ -438,10 +488,12 @@ static size_t ZSTD_compressSubBlock_multi(const seqStore_t* seqStorePtr,
 {
     const seqDef* const sstart = seqStorePtr->sequencesStart;
     const seqDef* const send = seqStorePtr->sequences;
-    const seqDef* sp = sstart;
+    const seqDef* sp = sstart; /* tracks progresses within seqStorePtr->sequences */
+    size_t const nbSeqs = (size_t)(send - sstart);
     const BYTE* const lstart = seqStorePtr->litStart;
     const BYTE* const lend = seqStorePtr->lit;
     const BYTE* lp = lstart;
+    size_t const nbLiterals = (size_t)(lend - lstart);
     BYTE const* ip = (BYTE const*)src;
     BYTE const* const iend = ip + srcSize;
     BYTE* const ostart = (BYTE*)dst;
@@ -450,96 +502,152 @@ static size_t ZSTD_compressSubBlock_multi(const seqStore_t* seqStorePtr,
     const BYTE* llCodePtr = seqStorePtr->llCode;
     const BYTE* mlCodePtr = seqStorePtr->mlCode;
     const BYTE* ofCodePtr = seqStorePtr->ofCode;
-    size_t targetCBlockSize = cctxParams->targetCBlockSize;
-    size_t litSize, seqCount;
-    int writeLitEntropy = entropyMetadata->hufMetadata.hType == set_compressed;
+    size_t const minTarget = ZSTD_TARGETCBLOCKSIZE_MIN; /* enforce minimum size, to reduce undesirable side effects */
+    size_t const targetCBlockSize = MAX(minTarget, cctxParams->targetCBlockSize);
+    int writeLitEntropy = (entropyMetadata->hufMetadata.hType == set_compressed);
     int writeSeqEntropy = 1;
-    int lastSequence = 0;
 
-    DEBUGLOG(5, "ZSTD_compressSubBlock_multi (litSize=%u, nbSeq=%u)",
-                (unsigned)(lend-lp), (unsigned)(send-sstart));
+    DEBUGLOG(5, "ZSTD_compressSubBlock_multi (srcSize=%u, litSize=%u, nbSeq=%u)",
+               (unsigned)srcSize, (unsigned)(lend-lstart), (unsigned)(send-sstart));
 
-    litSize = 0;
-    seqCount = 0;
-    do {
-        size_t cBlockSizeEstimate = 0;
-        if (sstart == send) {
-            lastSequence = 1;
-        } else {
-            const seqDef* const sequence = sp + seqCount;
-            lastSequence = sequence == send - 1;
-            litSize += ZSTD_getSequenceLength(seqStorePtr, sequence).litLength;
-            seqCount++;
+        /* let's start by a general estimation for the full block */
+    if (nbSeqs > 0) {
+        EstimatedBlockSize const ebs =
+                ZSTD_estimateSubBlockSize(lp, nbLiterals,
+                                        ofCodePtr, llCodePtr, mlCodePtr, nbSeqs,
+                                        &nextCBlock->entropy, entropyMetadata,
+                                        workspace, wkspSize,
+                                        writeLitEntropy, writeSeqEntropy);
+        /* quick estimation */
+        size_t const avgLitCost = nbLiterals ? (ebs.estLitSize * BYTESCALE) / nbLiterals : BYTESCALE;
+        size_t const avgSeqCost = ((ebs.estBlockSize - ebs.estLitSize) * BYTESCALE) / nbSeqs;
+        const size_t nbSubBlocks = MAX((ebs.estBlockSize + (targetCBlockSize/2)) / targetCBlockSize, 1);
+        size_t n, avgBlockBudget, blockBudgetSupp=0;
+        avgBlockBudget = (ebs.estBlockSize * BYTESCALE) / nbSubBlocks;
+        DEBUGLOG(5, "estimated fullblock size=%u bytes ; avgLitCost=%.2f ; avgSeqCost=%.2f ; targetCBlockSize=%u, nbSubBlocks=%u ; avgBlockBudget=%.0f bytes",
+                    (unsigned)ebs.estBlockSize, (double)avgLitCost/BYTESCALE, (double)avgSeqCost/BYTESCALE,
+                    (unsigned)targetCBlockSize, (unsigned)nbSubBlocks, (double)avgBlockBudget/BYTESCALE);
+        /* simplification: if estimates states that the full superblock doesn't compress, just bail out immediately
+         * this will result in the production of a single uncompressed block covering @srcSize.*/
+        if (ebs.estBlockSize > srcSize) return 0;
+
+        /* compress and write sub-blocks */
+        assert(nbSubBlocks>0);
+        for (n=0; n < nbSubBlocks-1; n++) {
+            /* determine nb of sequences for current sub-block + nbLiterals from next sequence */
+            size_t const seqCount = sizeBlockSequences(sp, (size_t)(send-sp),
+                                        avgBlockBudget + blockBudgetSupp, avgLitCost, avgSeqCost, n==0);
+            /* if reached last sequence : break to last sub-block (simplification) */
+            assert(seqCount <= (size_t)(send-sp));
+            if (sp + seqCount == send) break;
+            assert(seqCount > 0);
+            /* compress sub-block */
+            {   int litEntropyWritten = 0;
+                int seqEntropyWritten = 0;
+                size_t litSize = countLiterals(seqStorePtr, sp, seqCount);
+                const size_t decompressedSize =
+                        ZSTD_seqDecompressedSize(seqStorePtr, sp, seqCount, litSize, 0);
+                size_t const cSize = ZSTD_compressSubBlock(&nextCBlock->entropy, entropyMetadata,
+                                                sp, seqCount,
+                                                lp, litSize,
+                                                llCodePtr, mlCodePtr, ofCodePtr,
+                                                cctxParams,
+                                                op, (size_t)(oend-op),
+                                                bmi2, writeLitEntropy, writeSeqEntropy,
+                                                &litEntropyWritten, &seqEntropyWritten,
+                                                0);
+                FORWARD_IF_ERROR(cSize, "ZSTD_compressSubBlock failed");
+
+                /* check compressibility, update state components */
+                if (cSize > 0 && cSize < decompressedSize) {
+                    DEBUGLOG(5, "Committed sub-block compressing %u bytes => %u bytes",
+                                (unsigned)decompressedSize, (unsigned)cSize);
+                    assert(ip + decompressedSize <= iend);
+                    ip += decompressedSize;
+                    lp += litSize;
+                    op += cSize;
+                    llCodePtr += seqCount;
+                    mlCodePtr += seqCount;
+                    ofCodePtr += seqCount;
+                    /* Entropy only needs to be written once */
+                    if (litEntropyWritten) {
+                        writeLitEntropy = 0;
+                    }
+                    if (seqEntropyWritten) {
+                        writeSeqEntropy = 0;
+                    }
+                    sp += seqCount;
+                    blockBudgetSupp = 0;
+            }   }
+            /* otherwise : do not compress yet, coalesce current sub-block with following one */
         }
-        if (lastSequence) {
-            assert(lp <= lend);
-            assert(litSize <= (size_t)(lend - lp));
-            litSize = (size_t)(lend - lp);
-        }
-        /* I think there is an optimization opportunity here.
-         * Calling ZSTD_estimateSubBlockSize for every sequence can be wasteful
-         * since it recalculates estimate from scratch.
-         * For example, it would recount literal distribution and symbol codes every time.
-         */
-        cBlockSizeEstimate = ZSTD_estimateSubBlockSize(lp, litSize, ofCodePtr, llCodePtr, mlCodePtr, seqCount,
-                                                       &nextCBlock->entropy, entropyMetadata,
-                                                       workspace, wkspSize, writeLitEntropy, writeSeqEntropy);
-        if (cBlockSizeEstimate > targetCBlockSize || lastSequence) {
-            int litEntropyWritten = 0;
-            int seqEntropyWritten = 0;
-            const size_t decompressedSize = ZSTD_seqDecompressedSize(seqStorePtr, sp, seqCount, litSize, lastSequence);
-            const size_t cSize = ZSTD_compressSubBlock(&nextCBlock->entropy, entropyMetadata,
-                                                       sp, seqCount,
-                                                       lp, litSize,
-                                                       llCodePtr, mlCodePtr, ofCodePtr,
-                                                       cctxParams,
-                                                       op, oend-op,
-                                                       bmi2, writeLitEntropy, writeSeqEntropy,
-                                                       &litEntropyWritten, &seqEntropyWritten,
-                                                       lastBlock && lastSequence);
-            FORWARD_IF_ERROR(cSize, "ZSTD_compressSubBlock failed");
-            if (cSize > 0 && cSize < decompressedSize) {
-                DEBUGLOG(5, "Committed the sub-block");
-                assert(ip + decompressedSize <= iend);
-                ip += decompressedSize;
-                sp += seqCount;
-                lp += litSize;
-                op += cSize;
-                llCodePtr += seqCount;
-                mlCodePtr += seqCount;
-                ofCodePtr += seqCount;
-                litSize = 0;
-                seqCount = 0;
-                /* Entropy only needs to be written once */
-                if (litEntropyWritten) {
-                    writeLitEntropy = 0;
-                }
-                if (seqEntropyWritten) {
-                    writeSeqEntropy = 0;
-                }
+    } /* if (nbSeqs > 0) */
+
+    /* write last block */
+    DEBUGLOG(5, "Generate last sub-block: %u sequences remaining", (unsigned)(send - sp));
+    {   int litEntropyWritten = 0;
+        int seqEntropyWritten = 0;
+        size_t litSize = (size_t)(lend - lp);
+        size_t seqCount = (size_t)(send - sp);
+        const size_t decompressedSize =
+                ZSTD_seqDecompressedSize(seqStorePtr, sp, seqCount, litSize, 1);
+        size_t const cSize = ZSTD_compressSubBlock(&nextCBlock->entropy, entropyMetadata,
+                                            sp, seqCount,
+                                            lp, litSize,
+                                            llCodePtr, mlCodePtr, ofCodePtr,
+                                            cctxParams,
+                                            op, (size_t)(oend-op),
+                                            bmi2, writeLitEntropy, writeSeqEntropy,
+                                            &litEntropyWritten, &seqEntropyWritten,
+                                            lastBlock);
+        FORWARD_IF_ERROR(cSize, "ZSTD_compressSubBlock failed");
+
+        /* update pointers, the nb of literals borrowed from next sequence must be preserved */
+        if (cSize > 0 && cSize < decompressedSize) {
+            DEBUGLOG(5, "Last sub-block compressed %u bytes => %u bytes",
+                        (unsigned)decompressedSize, (unsigned)cSize);
+            assert(ip + decompressedSize <= iend);
+            ip += decompressedSize;
+            lp += litSize;
+            op += cSize;
+            llCodePtr += seqCount;
+            mlCodePtr += seqCount;
+            ofCodePtr += seqCount;
+            /* Entropy only needs to be written once */
+            if (litEntropyWritten) {
+                writeLitEntropy = 0;
             }
+            if (seqEntropyWritten) {
+                writeSeqEntropy = 0;
+            }
+            sp += seqCount;
         }
-    } while (!lastSequence);
+    }
+
+
     if (writeLitEntropy) {
-        DEBUGLOG(5, "ZSTD_compressSubBlock_multi has literal entropy tables unwritten");
+        DEBUGLOG(5, "Literal entropy tables were never written");
         ZSTD_memcpy(&nextCBlock->entropy.huf, &prevCBlock->entropy.huf, sizeof(prevCBlock->entropy.huf));
     }
     if (writeSeqEntropy && ZSTD_needSequenceEntropyTables(&entropyMetadata->fseMetadata)) {
         /* If we haven't written our entropy tables, then we've violated our contract and
          * must emit an uncompressed block.
          */
-        DEBUGLOG(5, "ZSTD_compressSubBlock_multi has sequence entropy tables unwritten");
+        DEBUGLOG(5, "Sequence entropy tables were never written => cancel, emit an uncompressed block");
         return 0;
     }
+
     if (ip < iend) {
-        size_t const cSize = ZSTD_noCompressBlock(op, oend - op, ip, iend - ip, lastBlock);
-        DEBUGLOG(5, "ZSTD_compressSubBlock_multi last sub-block uncompressed, %zu bytes", (size_t)(iend - ip));
+        /* some data left : last part of the block sent uncompressed */
+        size_t const rSize = (size_t)((iend - ip));
+        size_t const cSize = ZSTD_noCompressBlock(op, (size_t)(oend - op), ip, rSize, lastBlock);
+        DEBUGLOG(5, "Generate last uncompressed sub-block of %u bytes", (unsigned)(rSize));
         FORWARD_IF_ERROR(cSize, "ZSTD_noCompressBlock failed");
         assert(cSize != 0);
         op += cSize;
         /* We have to regenerate the repcodes because we've skipped some sequences */
         if (sp < send) {
-            seqDef const* seq;
+            const seqDef* seq;
             repcodes_t rep;
             ZSTD_memcpy(&rep, prevCBlock->rep, sizeof(rep));
             for (seq = sstart; seq < sp; ++seq) {
@@ -548,14 +656,17 @@ static size_t ZSTD_compressSubBlock_multi(const seqStore_t* seqStorePtr,
             ZSTD_memcpy(nextCBlock->rep, &rep, sizeof(rep));
         }
     }
-    DEBUGLOG(5, "ZSTD_compressSubBlock_multi compressed");
-    return op-ostart;
+
+    DEBUGLOG(5, "ZSTD_compressSubBlock_multi compressed all subBlocks: total compressed size = %u",
+                (unsigned)(op-ostart));
+    return (size_t)(op-ostart);
 }
 
 size_t ZSTD_compressSuperBlock(ZSTD_CCtx* zc,
                                void* dst, size_t dstCapacity,
-                               void const* src, size_t srcSize,
-                               unsigned lastBlock) {
+                               const void* src, size_t srcSize,
+                               unsigned lastBlock)
+{
     ZSTD_entropyCTablesMetadata_t entropyMetadata;
 
     FORWARD_IF_ERROR(ZSTD_buildBlockEntropyStats(&zc->seqStore,

lib/compress/zstd_cwksp.h

@@ -14,7 +14,9 @@
 /*-*************************************
 *  Dependencies
 ***************************************/
+#include "../common/allocations.h"  /* ZSTD_customMalloc, ZSTD_customFree */
 #include "../common/zstd_internal.h"
+#include "../common/portability_macros.h"
 
 #if defined (__cplusplus)
 extern "C" {
@@ -44,8 +46,9 @@ extern "C" {
 ***************************************/
 typedef enum {
     ZSTD_cwksp_alloc_objects,
-    ZSTD_cwksp_alloc_buffers,
-    ZSTD_cwksp_alloc_aligned
+    ZSTD_cwksp_alloc_aligned_init_once,
+    ZSTD_cwksp_alloc_aligned,
+    ZSTD_cwksp_alloc_buffers
 } ZSTD_cwksp_alloc_phase_e;
 
 /**
@@ -98,8 +101,8 @@ typedef enum {
  *
  * Workspace Layout:
  *
- * [                        ... workspace ...                         ]
- * [objects][tables ... ->] free space [<- ... aligned][<- ... buffers]
+ * [                        ... workspace ...                           ]
+ * [objects][tables ->] free space [<- buffers][<- aligned][<- init once]
  *
  * The various objects that live in the workspace are divided into the
  * following categories, and are allocated separately:
@@ -123,9 +126,18 @@ typedef enum {
  *   uint32_t arrays, all of whose values are between 0 and (nextSrc - base).
  *   Their sizes depend on the cparams. These tables are 64-byte aligned.
  *
- * - Aligned: these buffers are used for various purposes that require 4 byte
- *   alignment, but don't require any initialization before they're used. These
- *   buffers are each aligned to 64 bytes.
+ * - Init once: these buffers require to be initialized at least once before
+ *   use. They should be used when we want to skip memory initialization
+ *   while not triggering memory checkers (like Valgrind) when reading from
+ *   from this memory without writing to it first.
+ *   These buffers should be used carefully as they might contain data
+ *   from previous compressions.
+ *   Buffers are aligned to 64 bytes.
+ *
+ * - Aligned: these buffers don't require any initialization before they're
+ *   used. The user of the buffer should make sure they write into a buffer
+ *   location before reading from it.
+ *   Buffers are aligned to 64 bytes.
  *
  * - Buffers: these buffers are used for various purposes that don't require
  *   any alignment or initialization before they're used. This means they can
@@ -137,8 +149,9 @@ typedef enum {
  * correctly packed into the workspace buffer. That order is:
  *
  * 1. Objects
- * 2. Buffers
- * 3. Aligned/Tables
+ * 2. Init once / Tables
+ * 3. Aligned / Tables
+ * 4. Buffers / Tables
  *
  * Attempts to reserve objects of different types out of order will fail.
  */
@@ -150,6 +163,7 @@ typedef struct {
     void* tableEnd;
     void* tableValidEnd;
     void* allocStart;
+    void* initOnceStart;
 
     BYTE allocFailed;
     int workspaceOversizedDuration;
@@ -162,6 +176,7 @@ typedef struct {
 ***************************************/
 
 MEM_STATIC size_t ZSTD_cwksp_available_space(ZSTD_cwksp* ws);
+MEM_STATIC void*  ZSTD_cwksp_initialAllocStart(ZSTD_cwksp* ws);
 
 MEM_STATIC void ZSTD_cwksp_assert_internal_consistency(ZSTD_cwksp* ws) {
     (void)ws;
@@ -171,6 +186,21 @@ MEM_STATIC void ZSTD_cwksp_assert_internal_consistency(ZSTD_cwksp* ws) {
     assert(ws->tableEnd <= ws->allocStart);
     assert(ws->tableValidEnd <= ws->allocStart);
     assert(ws->allocStart <= ws->workspaceEnd);
+    assert(ws->initOnceStart <= ZSTD_cwksp_initialAllocStart(ws));
+    assert(ws->workspace <= ws->initOnceStart);
+#if ZSTD_MEMORY_SANITIZER
+    {
+        intptr_t const offset = __msan_test_shadow(ws->initOnceStart,
+            (U8*)ZSTD_cwksp_initialAllocStart(ws) - (U8*)ws->initOnceStart);
+        (void)offset;
+#if defined(ZSTD_MSAN_PRINT)
+        if(offset!=-1) {
+            __msan_print_shadow((U8*)ws->initOnceStart + offset - 8, 32);
+        }
+#endif
+        assert(offset==-1);
+    };
+#endif
 }
 
 /**
@@ -217,14 +247,10 @@ MEM_STATIC size_t ZSTD_cwksp_aligned_alloc_size(size_t size) {
  * for internal purposes (currently only alignment).
  */
 MEM_STATIC size_t ZSTD_cwksp_slack_space_required(void) {
-    /* For alignment, the wksp will always allocate an additional n_1=[1, 64] bytes
-     * to align the beginning of tables section, as well as another n_2=[0, 63] bytes
-     * to align the beginning of the aligned section.
-     *
-     * n_1 + n_2 == 64 bytes if the cwksp is freshly allocated, due to tables and
-     * aligneds being sized in multiples of 64 bytes.
+    /* For alignment, the wksp will always allocate an additional 2*ZSTD_CWKSP_ALIGNMENT_BYTES
+     * bytes to align the beginning of tables section and end of buffers;
      */
-    size_t const slackSpace = ZSTD_CWKSP_ALIGNMENT_BYTES;
+    size_t const slackSpace = ZSTD_CWKSP_ALIGNMENT_BYTES * 2;
     return slackSpace;
 }
 
@@ -237,10 +263,18 @@ MEM_STATIC size_t ZSTD_cwksp_bytes_to_align_ptr(void* ptr, const size_t alignByt
     size_t const alignBytesMask = alignBytes - 1;
     size_t const bytes = (alignBytes - ((size_t)ptr & (alignBytesMask))) & alignBytesMask;
     assert((alignBytes & alignBytesMask) == 0);
-    assert(bytes != ZSTD_CWKSP_ALIGNMENT_BYTES);
+    assert(bytes < alignBytes);
     return bytes;
 }
 
+/**
+ * Returns the initial value for allocStart which is used to determine the position from
+ * which we can allocate from the end of the workspace.
+ */
+MEM_STATIC void*  ZSTD_cwksp_initialAllocStart(ZSTD_cwksp* ws) {
+    return (void*)((size_t)ws->workspaceEnd & ~(ZSTD_CWKSP_ALIGNMENT_BYTES-1));
+}
+
 /**
  * Internal function. Do not use directly.
  * Reserves the given number of bytes within the aligned/buffer segment of the wksp,
@@ -281,27 +315,16 @@ ZSTD_cwksp_internal_advance_phase(ZSTD_cwksp* ws, ZSTD_cwksp_alloc_phase_e phase
 {
     assert(phase >= ws->phase);
     if (phase > ws->phase) {
-        /* Going from allocating objects to allocating buffers */
-        if (ws->phase < ZSTD_cwksp_alloc_buffers &&
-                phase >= ZSTD_cwksp_alloc_buffers) {
+        /* Going from allocating objects to allocating initOnce / tables */
+        if (ws->phase < ZSTD_cwksp_alloc_aligned_init_once &&
+            phase >= ZSTD_cwksp_alloc_aligned_init_once) {
             ws->tableValidEnd = ws->objectEnd;
-        }
+            ws->initOnceStart = ZSTD_cwksp_initialAllocStart(ws);
 
-        /* Going from allocating buffers to allocating aligneds/tables */
-        if (ws->phase < ZSTD_cwksp_alloc_aligned &&
-                phase >= ZSTD_cwksp_alloc_aligned) {
-            {   /* Align the start of the "aligned" to 64 bytes. Use [1, 64] bytes. */
-                size_t const bytesToAlign =
-                    ZSTD_CWKSP_ALIGNMENT_BYTES - ZSTD_cwksp_bytes_to_align_ptr(ws->allocStart, ZSTD_CWKSP_ALIGNMENT_BYTES);
-                DEBUGLOG(5, "reserving aligned alignment addtl space: %zu", bytesToAlign);
-                ZSTD_STATIC_ASSERT((ZSTD_CWKSP_ALIGNMENT_BYTES & (ZSTD_CWKSP_ALIGNMENT_BYTES - 1)) == 0); /* power of 2 */
-                RETURN_ERROR_IF(!ZSTD_cwksp_reserve_internal_buffer_space(ws, bytesToAlign),
-                                memory_allocation, "aligned phase - alignment initial allocation failed!");
-            }
             {   /* Align the start of the tables to 64 bytes. Use [0, 63] bytes */
-                void* const alloc = ws->objectEnd;
+                void *const alloc = ws->objectEnd;
                 size_t const bytesToAlign = ZSTD_cwksp_bytes_to_align_ptr(alloc, ZSTD_CWKSP_ALIGNMENT_BYTES);
-                void* const objectEnd = (BYTE*)alloc + bytesToAlign;
+                void *const objectEnd = (BYTE *) alloc + bytesToAlign;
                 DEBUGLOG(5, "reserving table alignment addtl space: %zu", bytesToAlign);
                 RETURN_ERROR_IF(objectEnd > ws->workspaceEnd, memory_allocation,
                                 "table phase - alignment initial allocation failed!");
@@ -309,7 +332,9 @@ ZSTD_cwksp_internal_advance_phase(ZSTD_cwksp* ws, ZSTD_cwksp_alloc_phase_e phase
                 ws->tableEnd = objectEnd;  /* table area starts being empty */
                 if (ws->tableValidEnd < ws->tableEnd) {
                     ws->tableValidEnd = ws->tableEnd;
-        }   }   }
+                }
+            }
+        }
         ws->phase = phase;
         ZSTD_cwksp_assert_internal_consistency(ws);
     }
@@ -321,7 +346,7 @@ ZSTD_cwksp_internal_advance_phase(ZSTD_cwksp* ws, ZSTD_cwksp_alloc_phase_e phase
  */
 MEM_STATIC int ZSTD_cwksp_owns_buffer(const ZSTD_cwksp* ws, const void* ptr)
 {
-    return (ptr != NULL) && (ws->workspace <= ptr) && (ptr <= ws->workspaceEnd);
+    return (ptr != NULL) && (ws->workspace <= ptr) && (ptr < ws->workspaceEnd);
 }
 
 /**
@@ -366,6 +391,36 @@ MEM_STATIC BYTE* ZSTD_cwksp_reserve_buffer(ZSTD_cwksp* ws, size_t bytes)
     return (BYTE*)ZSTD_cwksp_reserve_internal(ws, bytes, ZSTD_cwksp_alloc_buffers);
 }
 
+/**
+ * Reserves and returns memory sized on and aligned on ZSTD_CWKSP_ALIGNMENT_BYTES (64 bytes).
+ * This memory has been initialized at least once in the past.
+ * This doesn't mean it has been initialized this time, and it might contain data from previous
+ * operations.
+ * The main usage is for algorithms that might need read access into uninitialized memory.
+ * The algorithm must maintain safety under these conditions and must make sure it doesn't
+ * leak any of the past data (directly or in side channels).
+ */
+MEM_STATIC void* ZSTD_cwksp_reserve_aligned_init_once(ZSTD_cwksp* ws, size_t bytes)
+{
+    size_t const alignedBytes = ZSTD_cwksp_align(bytes, ZSTD_CWKSP_ALIGNMENT_BYTES);
+    void* ptr = ZSTD_cwksp_reserve_internal(ws, alignedBytes, ZSTD_cwksp_alloc_aligned_init_once);
+    assert(((size_t)ptr & (ZSTD_CWKSP_ALIGNMENT_BYTES-1))== 0);
+    if(ptr && ptr < ws->initOnceStart) {
+        /* We assume the memory following the current allocation is either:
+         * 1. Not usable as initOnce memory (end of workspace)
+         * 2. Another initOnce buffer that has been allocated before (and so was previously memset)
+         * 3. An ASAN redzone, in which case we don't want to write on it
+         * For these reasons it should be fine to not explicitly zero every byte up to ws->initOnceStart.
+         * Note that we assume here that MSAN and ASAN cannot run in the same time. */
+        ZSTD_memset(ptr, 0, MIN((size_t)((U8*)ws->initOnceStart - (U8*)ptr), alignedBytes));
+        ws->initOnceStart = ptr;
+    }
+#if ZSTD_MEMORY_SANITIZER
+    assert(__msan_test_shadow(ptr, bytes) == -1);
+#endif
+    return ptr;
+}
+
 /**
  * Reserves and returns memory sized on and aligned on ZSTD_CWKSP_ALIGNMENT_BYTES (64 bytes).
  */
@@ -379,18 +434,22 @@ MEM_STATIC void* ZSTD_cwksp_reserve_aligned(ZSTD_cwksp* ws, size_t bytes)
 
 /**
  * Aligned on 64 bytes. These buffers have the special property that
- * their values remain constrained, allowing us to re-use them without
+ * their values remain constrained, allowing us to reuse them without
  * memset()-ing them.
  */
 MEM_STATIC void* ZSTD_cwksp_reserve_table(ZSTD_cwksp* ws, size_t bytes)
 {
-    const ZSTD_cwksp_alloc_phase_e phase = ZSTD_cwksp_alloc_aligned;
+    const ZSTD_cwksp_alloc_phase_e phase = ZSTD_cwksp_alloc_aligned_init_once;
     void* alloc;
     void* end;
     void* top;
 
-    if (ZSTD_isError(ZSTD_cwksp_internal_advance_phase(ws, phase))) {
-        return NULL;
+    /* We can only start allocating tables after we are done reserving space for objects at the
+     * start of the workspace */
+    if(ws->phase < phase) {
+        if (ZSTD_isError(ZSTD_cwksp_internal_advance_phase(ws, phase))) {
+            return NULL;
+        }
     }
     alloc = ws->tableEnd;
     end = (BYTE *)alloc + bytes;
@@ -467,13 +526,21 @@ MEM_STATIC void ZSTD_cwksp_mark_tables_dirty(ZSTD_cwksp* ws)
     DEBUGLOG(4, "cwksp: ZSTD_cwksp_mark_tables_dirty");
 
 #if ZSTD_MEMORY_SANITIZER && !defined (ZSTD_MSAN_DONT_POISON_WORKSPACE)
-    /* To validate that the table re-use logic is sound, and that we don't
+    /* To validate that the table reuse logic is sound, and that we don't
      * access table space that we haven't cleaned, we re-"poison" the table
-     * space every time we mark it dirty. */
+     * space every time we mark it dirty.
+     * Since tableValidEnd space and initOnce space may overlap we don't poison
+     * the initOnce portion as it break its promise. This means that this poisoning
+     * check isn't always applied fully. */
     {
         size_t size = (BYTE*)ws->tableValidEnd - (BYTE*)ws->objectEnd;
         assert(__msan_test_shadow(ws->objectEnd, size) == -1);
-        __msan_poison(ws->objectEnd, size);
+        if((BYTE*)ws->tableValidEnd < (BYTE*)ws->initOnceStart) {
+            __msan_poison(ws->objectEnd, size);
+        } else {
+            assert(ws->initOnceStart >= ws->objectEnd);
+            __msan_poison(ws->objectEnd, (BYTE*)ws->initOnceStart - (BYTE*)ws->objectEnd);
+        }
     }
 #endif
 
@@ -536,13 +603,16 @@ MEM_STATIC void ZSTD_cwksp_clear(ZSTD_cwksp* ws) {
     DEBUGLOG(4, "cwksp: clearing!");
 
 #if ZSTD_MEMORY_SANITIZER && !defined (ZSTD_MSAN_DONT_POISON_WORKSPACE)
-    /* To validate that the context re-use logic is sound, and that we don't
+    /* To validate that the context reuse logic is sound, and that we don't
      * access stuff that this compression hasn't initialized, we re-"poison"
-     * the workspace (or at least the non-static, non-table parts of it)
-     * every time we start a new compression. */
+     * the workspace except for the areas in which we expect memory reuse
+     * without initialization (objects, valid tables area and init once
+     * memory). */
     {
-        size_t size = (BYTE*)ws->workspaceEnd - (BYTE*)ws->tableValidEnd;
-        __msan_poison(ws->tableValidEnd, size);
+        if((BYTE*)ws->tableValidEnd < (BYTE*)ws->initOnceStart) {
+            size_t size = (BYTE*)ws->initOnceStart - (BYTE*)ws->tableValidEnd;
+            __msan_poison(ws->tableValidEnd, size);
+        }
     }
 #endif
 
@@ -558,14 +628,23 @@ MEM_STATIC void ZSTD_cwksp_clear(ZSTD_cwksp* ws) {
 #endif
 
     ws->tableEnd = ws->objectEnd;
-    ws->allocStart = ws->workspaceEnd;
+    ws->allocStart = ZSTD_cwksp_initialAllocStart(ws);
     ws->allocFailed = 0;
-    if (ws->phase > ZSTD_cwksp_alloc_buffers) {
-        ws->phase = ZSTD_cwksp_alloc_buffers;
+    if (ws->phase > ZSTD_cwksp_alloc_aligned_init_once) {
+        ws->phase = ZSTD_cwksp_alloc_aligned_init_once;
     }
     ZSTD_cwksp_assert_internal_consistency(ws);
 }
 
+MEM_STATIC size_t ZSTD_cwksp_sizeof(const ZSTD_cwksp* ws) {
+    return (size_t)((BYTE*)ws->workspaceEnd - (BYTE*)ws->workspace);
+}
+
+MEM_STATIC size_t ZSTD_cwksp_used(const ZSTD_cwksp* ws) {
+    return (size_t)((BYTE*)ws->tableEnd - (BYTE*)ws->workspace)
+         + (size_t)((BYTE*)ws->workspaceEnd - (BYTE*)ws->allocStart);
+}
+
 /**
  * The provided workspace takes ownership of the buffer [start, start+size).
  * Any existing values in the workspace are ignored (the previously managed
@@ -578,6 +657,7 @@ MEM_STATIC void ZSTD_cwksp_init(ZSTD_cwksp* ws, void* start, size_t size, ZSTD_c
     ws->workspaceEnd = (BYTE*)start + size;
     ws->objectEnd = ws->workspace;
     ws->tableValidEnd = ws->objectEnd;
+    ws->initOnceStart = ZSTD_cwksp_initialAllocStart(ws);
     ws->phase = ZSTD_cwksp_alloc_objects;
     ws->isStatic = isStatic;
     ZSTD_cwksp_clear(ws);
@@ -596,6 +676,11 @@ MEM_STATIC size_t ZSTD_cwksp_create(ZSTD_cwksp* ws, size_t size, ZSTD_customMem
 MEM_STATIC void ZSTD_cwksp_free(ZSTD_cwksp* ws, ZSTD_customMem customMem) {
     void *ptr = ws->workspace;
     DEBUGLOG(4, "cwksp: freeing workspace");
+#if ZSTD_MEMORY_SANITIZER && !defined(ZSTD_MSAN_DONT_POISON_WORKSPACE)
+    if (ptr != NULL && customMem.customFree != NULL) {
+        __msan_unpoison(ptr, ZSTD_cwksp_sizeof(ws));
+    }
+#endif
     ZSTD_memset(ws, 0, sizeof(ZSTD_cwksp));
     ZSTD_customFree(ptr, customMem);
 }
@@ -609,15 +694,6 @@ MEM_STATIC void ZSTD_cwksp_move(ZSTD_cwksp* dst, ZSTD_cwksp* src) {
     ZSTD_memset(src, 0, sizeof(ZSTD_cwksp));
 }
 
-MEM_STATIC size_t ZSTD_cwksp_sizeof(const ZSTD_cwksp* ws) {
-    return (size_t)((BYTE*)ws->workspaceEnd - (BYTE*)ws->workspace);
-}
-
-MEM_STATIC size_t ZSTD_cwksp_used(const ZSTD_cwksp* ws) {
-    return (size_t)((BYTE*)ws->tableEnd - (BYTE*)ws->workspace)
-         + (size_t)((BYTE*)ws->workspaceEnd - (BYTE*)ws->allocStart);
-}
-
 MEM_STATIC int ZSTD_cwksp_reserve_failed(const ZSTD_cwksp* ws) {
     return ws->allocFailed;
 }
@@ -630,17 +706,11 @@ MEM_STATIC int ZSTD_cwksp_reserve_failed(const ZSTD_cwksp* ws) {
  * Returns if the estimated space needed for a wksp is within an acceptable limit of the
  * actual amount of space used.
  */
-MEM_STATIC int ZSTD_cwksp_estimated_space_within_bounds(const ZSTD_cwksp* const ws,
-                                                        size_t const estimatedSpace, int resizedWorkspace) {
-    if (resizedWorkspace) {
-        /* Resized/newly allocated wksp should have exact bounds */
-        return ZSTD_cwksp_used(ws) == estimatedSpace;
-    } else {
-        /* Due to alignment, when reusing a workspace, we can actually consume 63 fewer or more bytes
-         * than estimatedSpace. See the comments in zstd_cwksp.h for details.
-         */
-        return (ZSTD_cwksp_used(ws) >= estimatedSpace - 63) && (ZSTD_cwksp_used(ws) <= estimatedSpace + 63);
-    }
+MEM_STATIC int ZSTD_cwksp_estimated_space_within_bounds(const ZSTD_cwksp *const ws, size_t const estimatedSpace) {
+    /* We have an alignment space between objects and tables between tables and buffers, so we can have up to twice
+     * the alignment bytes difference between estimation and actual usage */
+    return (estimatedSpace - ZSTD_cwksp_slack_space_required()) <= ZSTD_cwksp_used(ws) &&
+           ZSTD_cwksp_used(ws) <= estimatedSpace;
 }
 
 

lib/compress/zstd_double_fast.c

@@ -11,7 +11,11 @@
 #include "zstd_compress_internal.h"
 #include "zstd_double_fast.h"
 
-static void ZSTD_fillDoubleHashTableForCDict(ZSTD_matchState_t* ms,
+#ifndef ZSTD_EXCLUDE_DFAST_BLOCK_COMPRESSOR
+
+static
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+void ZSTD_fillDoubleHashTableForCDict(ZSTD_matchState_t* ms,
                               void const* end, ZSTD_dictTableLoadMethod_e dtlm)
 {
     const ZSTD_compressionParameters* const cParams = &ms->cParams;
@@ -47,7 +51,9 @@ static void ZSTD_fillDoubleHashTableForCDict(ZSTD_matchState_t* ms,
     }   }
 }
 
-static void ZSTD_fillDoubleHashTableForCCtx(ZSTD_matchState_t* ms,
+static
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+void ZSTD_fillDoubleHashTableForCCtx(ZSTD_matchState_t* ms,
                               void const* end, ZSTD_dictTableLoadMethod_e dtlm)
 {
     const ZSTD_compressionParameters* const cParams = &ms->cParams;
@@ -95,6 +101,7 @@ void ZSTD_fillDoubleHashTable(ZSTD_matchState_t* ms,
 
 
 FORCE_INLINE_TEMPLATE
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
 size_t ZSTD_compressBlock_doubleFast_noDict_generic(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize, U32 const mls /* template */)
@@ -305,6 +312,7 @@ _match_stored:
 
 
 FORCE_INLINE_TEMPLATE
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
 size_t ZSTD_compressBlock_doubleFast_dictMatchState_generic(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize,
@@ -348,8 +356,8 @@ size_t ZSTD_compressBlock_doubleFast_dictMatchState_generic(
     if (ms->prefetchCDictTables) {
         size_t const hashTableBytes = (((size_t)1) << dictCParams->hashLog) * sizeof(U32);
         size_t const chainTableBytes = (((size_t)1) << dictCParams->chainLog) * sizeof(U32);
-        PREFETCH_AREA(dictHashLong, hashTableBytes)
-        PREFETCH_AREA(dictHashSmall, chainTableBytes)
+        PREFETCH_AREA(dictHashLong, hashTableBytes);
+        PREFETCH_AREA(dictHashSmall, chainTableBytes);
     }
 
     /* init */
@@ -589,7 +597,9 @@ size_t ZSTD_compressBlock_doubleFast_dictMatchState(
 }
 
 
-static size_t ZSTD_compressBlock_doubleFast_extDict_generic(
+static
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+size_t ZSTD_compressBlock_doubleFast_extDict_generic(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize,
         U32 const mls /* template */)
@@ -756,3 +766,5 @@ size_t ZSTD_compressBlock_doubleFast_extDict(
         return ZSTD_compressBlock_doubleFast_extDict_7(ms, seqStore, rep, src, srcSize);
     }
 }
+
+#endif /* ZSTD_EXCLUDE_DFAST_BLOCK_COMPRESSOR */

lib/compress/zstd_double_fast.h

@@ -18,9 +18,12 @@ extern "C" {
 #include "../common/mem.h"      /* U32 */
 #include "zstd_compress_internal.h"     /* ZSTD_CCtx, size_t */
 
+#ifndef ZSTD_EXCLUDE_DFAST_BLOCK_COMPRESSOR
+
 void ZSTD_fillDoubleHashTable(ZSTD_matchState_t* ms,
                               void const* end, ZSTD_dictTableLoadMethod_e dtlm,
                               ZSTD_tableFillPurpose_e tfp);
+
 size_t ZSTD_compressBlock_doubleFast(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
@@ -31,6 +34,14 @@ size_t ZSTD_compressBlock_doubleFast_extDict(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
 
+#define ZSTD_COMPRESSBLOCK_DOUBLEFAST ZSTD_compressBlock_doubleFast
+#define ZSTD_COMPRESSBLOCK_DOUBLEFAST_DICTMATCHSTATE ZSTD_compressBlock_doubleFast_dictMatchState
+#define ZSTD_COMPRESSBLOCK_DOUBLEFAST_EXTDICT ZSTD_compressBlock_doubleFast_extDict
+#else
+#define ZSTD_COMPRESSBLOCK_DOUBLEFAST NULL
+#define ZSTD_COMPRESSBLOCK_DOUBLEFAST_DICTMATCHSTATE NULL
+#define ZSTD_COMPRESSBLOCK_DOUBLEFAST_EXTDICT NULL
+#endif /* ZSTD_EXCLUDE_DFAST_BLOCK_COMPRESSOR */
 
 #if defined (__cplusplus)
 }

lib/compress/zstd_fast.c

@@ -11,7 +11,9 @@
 #include "zstd_compress_internal.h"  /* ZSTD_hashPtr, ZSTD_count, ZSTD_storeSeq */
 #include "zstd_fast.h"
 
-static void ZSTD_fillHashTableForCDict(ZSTD_matchState_t* ms,
+static
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+void ZSTD_fillHashTableForCDict(ZSTD_matchState_t* ms,
                         const void* const end,
                         ZSTD_dictTableLoadMethod_e dtlm)
 {
@@ -46,7 +48,9 @@ static void ZSTD_fillHashTableForCDict(ZSTD_matchState_t* ms,
                 }   }   }   }
 }
 
-static void ZSTD_fillHashTableForCCtx(ZSTD_matchState_t* ms,
+static
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+void ZSTD_fillHashTableForCCtx(ZSTD_matchState_t* ms,
                         const void* const end,
                         ZSTD_dictTableLoadMethod_e dtlm)
 {
@@ -139,8 +143,9 @@ void ZSTD_fillHashTable(ZSTD_matchState_t* ms,
  *
  * This is also the work we do at the beginning to enter the loop initially.
  */
-FORCE_INLINE_TEMPLATE size_t
-ZSTD_compressBlock_fast_noDict_generic(
+FORCE_INLINE_TEMPLATE
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+size_t ZSTD_compressBlock_fast_noDict_generic(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize,
         U32 const mls, U32 const hasStep)
@@ -456,6 +461,7 @@ size_t ZSTD_compressBlock_fast(
 }
 
 FORCE_INLINE_TEMPLATE
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
 size_t ZSTD_compressBlock_fast_dictMatchState_generic(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize, U32 const mls, U32 const hasStep)
@@ -502,7 +508,7 @@ size_t ZSTD_compressBlock_fast_dictMatchState_generic(
 
     if (ms->prefetchCDictTables) {
         size_t const hashTableBytes = (((size_t)1) << dictCParams->hashLog) * sizeof(U32);
-        PREFETCH_AREA(dictHashTable, hashTableBytes)
+        PREFETCH_AREA(dictHashTable, hashTableBytes);
     }
 
     /* init */
@@ -681,7 +687,9 @@ size_t ZSTD_compressBlock_fast_dictMatchState(
 }
 
 
-static size_t ZSTD_compressBlock_fast_extDict_generic(
+static
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+size_t ZSTD_compressBlock_fast_extDict_generic(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize, U32 const mls, U32 const hasStep)
 {

lib/compress/zstd_lazy.c

@@ -12,13 +12,21 @@
 #include "zstd_lazy.h"
 #include "../common/bits.h" /* ZSTD_countTrailingZeros64 */
 
+#if !defined(ZSTD_EXCLUDE_GREEDY_BLOCK_COMPRESSOR) \
+ || !defined(ZSTD_EXCLUDE_LAZY_BLOCK_COMPRESSOR) \
+ || !defined(ZSTD_EXCLUDE_LAZY2_BLOCK_COMPRESSOR) \
+ || !defined(ZSTD_EXCLUDE_BTLAZY2_BLOCK_COMPRESSOR)
+
+#define kLazySkippingStep 8
+
 
 /*-*************************************
 *  Binary Tree search
 ***************************************/
 
-static void
-ZSTD_updateDUBT(ZSTD_matchState_t* ms,
+static
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+void ZSTD_updateDUBT(ZSTD_matchState_t* ms,
                 const BYTE* ip, const BYTE* iend,
                 U32 mls)
 {
@@ -61,8 +69,9 @@ ZSTD_updateDUBT(ZSTD_matchState_t* ms,
  *  sort one already inserted but unsorted position
  *  assumption : curr >= btlow == (curr - btmask)
  *  doesn't fail */
-static void
-ZSTD_insertDUBT1(const ZSTD_matchState_t* ms,
+static
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+void ZSTD_insertDUBT1(const ZSTD_matchState_t* ms,
                  U32 curr, const BYTE* inputEnd,
                  U32 nbCompares, U32 btLow,
                  const ZSTD_dictMode_e dictMode)
@@ -150,8 +159,9 @@ ZSTD_insertDUBT1(const ZSTD_matchState_t* ms,
 }
 
 
-static size_t
-ZSTD_DUBT_findBetterDictMatch (
+static
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+size_t ZSTD_DUBT_findBetterDictMatch (
         const ZSTD_matchState_t* ms,
         const BYTE* const ip, const BYTE* const iend,
         size_t* offsetPtr,
@@ -228,8 +238,9 @@ ZSTD_DUBT_findBetterDictMatch (
 }
 
 
-static size_t
-ZSTD_DUBT_findBestMatch(ZSTD_matchState_t* ms,
+static
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+size_t ZSTD_DUBT_findBestMatch(ZSTD_matchState_t* ms,
                         const BYTE* const ip, const BYTE* const iend,
                         size_t* offBasePtr,
                         U32 const mls,
@@ -379,8 +390,9 @@ ZSTD_DUBT_findBestMatch(ZSTD_matchState_t* ms,
 
 
 /** ZSTD_BtFindBestMatch() : Tree updater, providing best match */
-FORCE_INLINE_TEMPLATE size_t
-ZSTD_BtFindBestMatch( ZSTD_matchState_t* ms,
+FORCE_INLINE_TEMPLATE
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+size_t ZSTD_BtFindBestMatch( ZSTD_matchState_t* ms,
                 const BYTE* const ip, const BYTE* const iLimit,
                       size_t* offBasePtr,
                 const U32 mls /* template */,
@@ -615,10 +627,12 @@ size_t ZSTD_dedicatedDictSearch_lazy_search(size_t* offsetPtr, size_t ml, U32 nb
 
 /* Update chains up to ip (excluded)
    Assumption : always within prefix (i.e. not within extDict) */
-FORCE_INLINE_TEMPLATE U32 ZSTD_insertAndFindFirstIndex_internal(
+FORCE_INLINE_TEMPLATE
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+U32 ZSTD_insertAndFindFirstIndex_internal(
                         ZSTD_matchState_t* ms,
                         const ZSTD_compressionParameters* const cParams,
-                        const BYTE* ip, U32 const mls)
+                        const BYTE* ip, U32 const mls, U32 const lazySkipping)
 {
     U32* const hashTable  = ms->hashTable;
     const U32 hashLog = cParams->hashLog;
@@ -633,6 +647,9 @@ FORCE_INLINE_TEMPLATE U32 ZSTD_insertAndFindFirstIndex_internal(
         NEXT_IN_CHAIN(idx, chainMask) = hashTable[h];
         hashTable[h] = idx;
         idx++;
+        /* Stop inserting every position when in the lazy skipping mode. */
+        if (lazySkipping)
+            break;
     }
 
     ms->nextToUpdate = target;
@@ -641,11 +658,12 @@ FORCE_INLINE_TEMPLATE U32 ZSTD_insertAndFindFirstIndex_internal(
 
 U32 ZSTD_insertAndFindFirstIndex(ZSTD_matchState_t* ms, const BYTE* ip) {
     const ZSTD_compressionParameters* const cParams = &ms->cParams;
-    return ZSTD_insertAndFindFirstIndex_internal(ms, cParams, ip, ms->cParams.minMatch);
+    return ZSTD_insertAndFindFirstIndex_internal(ms, cParams, ip, ms->cParams.minMatch, /* lazySkipping*/ 0);
 }
 
 /* inlining is important to hardwire a hot branch (template emulation) */
 FORCE_INLINE_TEMPLATE
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
 size_t ZSTD_HcFindBestMatch(
                         ZSTD_matchState_t* ms,
                         const BYTE* const ip, const BYTE* const iLimit,
@@ -685,7 +703,7 @@ size_t ZSTD_HcFindBestMatch(
     }
 
     /* HC4 match finder */
-    matchIndex = ZSTD_insertAndFindFirstIndex_internal(ms, cParams, ip, mls);
+    matchIndex = ZSTD_insertAndFindFirstIndex_internal(ms, cParams, ip, mls, ms->lazySkipping);
 
     for ( ; (matchIndex>=lowLimit) & (nbAttempts>0) ; nbAttempts--) {
         size_t currentMl=0;
@@ -758,7 +776,6 @@ size_t ZSTD_HcFindBestMatch(
 * (SIMD) Row-based matchfinder
 ***********************************/
 /* Constants for row-based hash */
-#define ZSTD_ROW_HASH_TAG_OFFSET 16     /* byte offset of hashes in the match state's tagTable from the beginning of a row */
 #define ZSTD_ROW_HASH_TAG_MASK ((1u << ZSTD_ROW_HASH_TAG_BITS) - 1)
 #define ZSTD_ROW_HASH_MAX_ENTRIES 64    /* absolute maximum number of entries per row, for all configurations */
 
@@ -774,39 +791,15 @@ MEM_STATIC U32 ZSTD_VecMask_next(ZSTD_VecMask val) {
     return ZSTD_countTrailingZeros64(val);
 }
 
-/* ZSTD_rotateRight_*():
- * Rotates a bitfield to the right by "count" bits.
- * https://en.wikipedia.org/w/index.php?title=Circular_shift&oldid=991635599#Implementing_circular_shifts
- */
-FORCE_INLINE_TEMPLATE
-U64 ZSTD_rotateRight_U64(U64 const value, U32 count) {
-    assert(count < 64);
-    count &= 0x3F; /* for fickle pattern recognition */
-    return (value >> count) | (U64)(value << ((0U - count) & 0x3F));
-}
-
-FORCE_INLINE_TEMPLATE
-U32 ZSTD_rotateRight_U32(U32 const value, U32 count) {
-    assert(count < 32);
-    count &= 0x1F; /* for fickle pattern recognition */
-    return (value >> count) | (U32)(value << ((0U - count) & 0x1F));
-}
-
-FORCE_INLINE_TEMPLATE
-U16 ZSTD_rotateRight_U16(U16 const value, U32 count) {
-    assert(count < 16);
-    count &= 0x0F; /* for fickle pattern recognition */
-    return (value >> count) | (U16)(value << ((0U - count) & 0x0F));
-}
-
 /* ZSTD_row_nextIndex():
  * Returns the next index to insert at within a tagTable row, and updates the "head"
- * value to reflect the update. Essentially cycles backwards from [0, {entries per row})
+ * value to reflect the update. Essentially cycles backwards from [1, {entries per row})
  */
 FORCE_INLINE_TEMPLATE U32 ZSTD_row_nextIndex(BYTE* const tagRow, U32 const rowMask) {
-  U32 const next = (*tagRow - 1) & rowMask;
-  *tagRow = (BYTE)next;
-  return next;
+    U32 next = (*tagRow-1) & rowMask;
+    next += (next == 0) ? rowMask : 0; /* skip first position */
+    *tagRow = (BYTE)next;
+    return next;
 }
 
 /* ZSTD_isAligned():
@@ -820,7 +813,7 @@ MEM_STATIC int ZSTD_isAligned(void const* ptr, size_t align) {
 /* ZSTD_row_prefetch():
  * Performs prefetching for the hashTable and tagTable at a given row.
  */
-FORCE_INLINE_TEMPLATE void ZSTD_row_prefetch(U32 const* hashTable, U16 const* tagTable, U32 const relRow, U32 const rowLog) {
+FORCE_INLINE_TEMPLATE void ZSTD_row_prefetch(U32 const* hashTable, BYTE const* tagTable, U32 const relRow, U32 const rowLog) {
     PREFETCH_L1(hashTable + relRow);
     if (rowLog >= 5) {
         PREFETCH_L1(hashTable + relRow + 16);
@@ -839,18 +832,20 @@ FORCE_INLINE_TEMPLATE void ZSTD_row_prefetch(U32 const* hashTable, U16 const* ta
  * Fill up the hash cache starting at idx, prefetching up to ZSTD_ROW_HASH_CACHE_SIZE entries,
  * but not beyond iLimit.
  */
-FORCE_INLINE_TEMPLATE void ZSTD_row_fillHashCache(ZSTD_matchState_t* ms, const BYTE* base,
+FORCE_INLINE_TEMPLATE
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+void ZSTD_row_fillHashCache(ZSTD_matchState_t* ms, const BYTE* base,
                                    U32 const rowLog, U32 const mls,
                                    U32 idx, const BYTE* const iLimit)
 {
     U32 const* const hashTable = ms->hashTable;
-    U16 const* const tagTable = ms->tagTable;
+    BYTE const* const tagTable = ms->tagTable;
     U32 const hashLog = ms->rowHashLog;
     U32 const maxElemsToPrefetch = (base + idx) > iLimit ? 0 : (U32)(iLimit - (base + idx) + 1);
     U32 const lim = idx + MIN(ZSTD_ROW_HASH_CACHE_SIZE, maxElemsToPrefetch);
 
     for (; idx < lim; ++idx) {
-        U32 const hash = (U32)ZSTD_hashPtr(base + idx, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls);
+        U32 const hash = (U32)ZSTD_hashPtrSalted(base + idx, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls, ms->hashSalt);
         U32 const row = (hash >> ZSTD_ROW_HASH_TAG_BITS) << rowLog;
         ZSTD_row_prefetch(hashTable, tagTable, row, rowLog);
         ms->hashCache[idx & ZSTD_ROW_HASH_CACHE_MASK] = hash;
@@ -865,12 +860,15 @@ FORCE_INLINE_TEMPLATE void ZSTD_row_fillHashCache(ZSTD_matchState_t* ms, const B
  * Returns the hash of base + idx, and replaces the hash in the hash cache with the byte at
  * base + idx + ZSTD_ROW_HASH_CACHE_SIZE. Also prefetches the appropriate rows from hashTable and tagTable.
  */
-FORCE_INLINE_TEMPLATE U32 ZSTD_row_nextCachedHash(U32* cache, U32 const* hashTable,
-                                                  U16 const* tagTable, BYTE const* base,
+FORCE_INLINE_TEMPLATE
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+U32 ZSTD_row_nextCachedHash(U32* cache, U32 const* hashTable,
+                                                  BYTE const* tagTable, BYTE const* base,
                                                   U32 idx, U32 const hashLog,
-                                                  U32 const rowLog, U32 const mls)
+                                                  U32 const rowLog, U32 const mls,
+                                                  U64 const hashSalt)
 {
-    U32 const newHash = (U32)ZSTD_hashPtr(base+idx+ZSTD_ROW_HASH_CACHE_SIZE, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls);
+    U32 const newHash = (U32)ZSTD_hashPtrSalted(base+idx+ZSTD_ROW_HASH_CACHE_SIZE, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls, hashSalt);
     U32 const row = (newHash >> ZSTD_ROW_HASH_TAG_BITS) << rowLog;
     ZSTD_row_prefetch(hashTable, tagTable, row, rowLog);
     {   U32 const hash = cache[idx & ZSTD_ROW_HASH_CACHE_MASK];
@@ -882,28 +880,29 @@ FORCE_INLINE_TEMPLATE U32 ZSTD_row_nextCachedHash(U32* cache, U32 const* hashTab
 /* ZSTD_row_update_internalImpl():
  * Updates the hash table with positions starting from updateStartIdx until updateEndIdx.
  */
-FORCE_INLINE_TEMPLATE void ZSTD_row_update_internalImpl(ZSTD_matchState_t* ms,
-                                                        U32 updateStartIdx, U32 const updateEndIdx,
-                                                        U32 const mls, U32 const rowLog,
-                                                        U32 const rowMask, U32 const useCache)
+FORCE_INLINE_TEMPLATE
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+void ZSTD_row_update_internalImpl(ZSTD_matchState_t* ms,
+                                  U32 updateStartIdx, U32 const updateEndIdx,
+                                  U32 const mls, U32 const rowLog,
+                                  U32 const rowMask, U32 const useCache)
 {
     U32* const hashTable = ms->hashTable;
-    U16* const tagTable = ms->tagTable;
+    BYTE* const tagTable = ms->tagTable;
     U32 const hashLog = ms->rowHashLog;
     const BYTE* const base = ms->window.base;
 
     DEBUGLOG(6, "ZSTD_row_update_internalImpl(): updateStartIdx=%u, updateEndIdx=%u", updateStartIdx, updateEndIdx);
     for (; updateStartIdx < updateEndIdx; ++updateStartIdx) {
-        U32 const hash = useCache ? ZSTD_row_nextCachedHash(ms->hashCache, hashTable, tagTable, base, updateStartIdx, hashLog, rowLog, mls)
-                                  : (U32)ZSTD_hashPtr(base + updateStartIdx, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls);
+        U32 const hash = useCache ? ZSTD_row_nextCachedHash(ms->hashCache, hashTable, tagTable, base, updateStartIdx, hashLog, rowLog, mls, ms->hashSalt)
+                                  : (U32)ZSTD_hashPtrSalted(base + updateStartIdx, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls, ms->hashSalt);
         U32 const relRow = (hash >> ZSTD_ROW_HASH_TAG_BITS) << rowLog;
         U32* const row = hashTable + relRow;
-        BYTE* tagRow = (BYTE*)(tagTable + relRow);  /* Though tagTable is laid out as a table of U16, each tag is only 1 byte.
-                                                       Explicit cast allows us to get exact desired position within each row */
+        BYTE* tagRow = tagTable + relRow;
         U32 const pos = ZSTD_row_nextIndex(tagRow, rowMask);
 
-        assert(hash == ZSTD_hashPtr(base + updateStartIdx, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls));
-        ((BYTE*)tagRow)[pos + ZSTD_ROW_HASH_TAG_OFFSET] = hash & ZSTD_ROW_HASH_TAG_MASK;
+        assert(hash == ZSTD_hashPtrSalted(base + updateStartIdx, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls, ms->hashSalt));
+        tagRow[pos] = hash & ZSTD_ROW_HASH_TAG_MASK;
         row[pos] = updateStartIdx;
     }
 }
@@ -912,9 +911,11 @@ FORCE_INLINE_TEMPLATE void ZSTD_row_update_internalImpl(ZSTD_matchState_t* ms,
  * Inserts the byte at ip into the appropriate position in the hash table, and updates ms->nextToUpdate.
  * Skips sections of long matches as is necessary.
  */
-FORCE_INLINE_TEMPLATE void ZSTD_row_update_internal(ZSTD_matchState_t* ms, const BYTE* ip,
-                                                    U32 const mls, U32 const rowLog,
-                                                    U32 const rowMask, U32 const useCache)
+FORCE_INLINE_TEMPLATE
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+void ZSTD_row_update_internal(ZSTD_matchState_t* ms, const BYTE* ip,
+                              U32 const mls, U32 const rowLog,
+                              U32 const rowMask, U32 const useCache)
 {
     U32 idx = ms->nextToUpdate;
     const BYTE* const base = ms->window.base;
@@ -1059,7 +1060,7 @@ ZSTD_row_getNEONMask(const U32 rowEntries, const BYTE* const src, const BYTE tag
 FORCE_INLINE_TEMPLATE ZSTD_VecMask
 ZSTD_row_getMatchMask(const BYTE* const tagRow, const BYTE tag, const U32 headGrouped, const U32 rowEntries)
 {
-    const BYTE* const src = tagRow + ZSTD_ROW_HASH_TAG_OFFSET;
+    const BYTE* const src = tagRow;
     assert((rowEntries == 16) || (rowEntries == 32) || rowEntries == 64);
     assert(rowEntries <= ZSTD_ROW_HASH_MAX_ENTRIES);
     assert(ZSTD_row_matchMaskGroupWidth(rowEntries) * rowEntries <= sizeof(ZSTD_VecMask) * 8);
@@ -1122,20 +1123,21 @@ ZSTD_row_getMatchMask(const BYTE* const tagRow, const BYTE tag, const U32 headGr
 
 /* The high-level approach of the SIMD row based match finder is as follows:
  * - Figure out where to insert the new entry:
- *      - Generate a hash from a byte along with an additional 1-byte "short hash". The additional byte is our "tag"
- *      - The hashTable is effectively split into groups or "rows" of 16 or 32 entries of U32, and the hash determines
+ *      - Generate a hash for current input posistion and split it into a one byte of tag and `rowHashLog` bits of index.
+ *           - The hash is salted by a value that changes on every contex reset, so when the same table is used
+ *             we will avoid collisions that would otherwise slow us down by intorducing phantom matches.
+ *      - The hashTable is effectively split into groups or "rows" of 15 or 31 entries of U32, and the index determines
  *        which row to insert into.
- *      - Determine the correct position within the row to insert the entry into. Each row of 16 or 32 can
- *        be considered as a circular buffer with a "head" index that resides in the tagTable.
- *      - Also insert the "tag" into the equivalent row and position in the tagTable.
- *          - Note: The tagTable has 17 or 33 1-byte entries per row, due to 16 or 32 tags, and 1 "head" entry.
- *                  The 17 or 33 entry rows are spaced out to occur every 32 or 64 bytes, respectively,
- *                  for alignment/performance reasons, leaving some bytes unused.
- * - Use SIMD to efficiently compare the tags in the tagTable to the 1-byte "short hash" and
+ *      - Determine the correct position within the row to insert the entry into. Each row of 15 or 31 can
+ *        be considered as a circular buffer with a "head" index that resides in the tagTable (overall 16 or 32 bytes
+ *        per row).
+ * - Use SIMD to efficiently compare the tags in the tagTable to the 1-byte tag calculated for the position and
  *   generate a bitfield that we can cycle through to check the collisions in the hash table.
  * - Pick the longest match.
+ * - Insert the tag into the equivalent row and position in the tagTable.
  */
 FORCE_INLINE_TEMPLATE
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
 size_t ZSTD_RowFindBestMatch(
                         ZSTD_matchState_t* ms,
                         const BYTE* const ip, const BYTE* const iLimit,
@@ -1144,7 +1146,7 @@ size_t ZSTD_RowFindBestMatch(
                         const U32 rowLog)
 {
     U32* const hashTable = ms->hashTable;
-    U16* const tagTable = ms->tagTable;
+    BYTE* const tagTable = ms->tagTable;
     U32* const hashCache = ms->hashCache;
     const U32 hashLog = ms->rowHashLog;
     const ZSTD_compressionParameters* const cParams = &ms->cParams;
@@ -1163,8 +1165,10 @@ size_t ZSTD_RowFindBestMatch(
     const U32 rowMask = rowEntries - 1;
     const U32 cappedSearchLog = MIN(cParams->searchLog, rowLog); /* nb of searches is capped at nb entries per row */
     const U32 groupWidth = ZSTD_row_matchMaskGroupWidth(rowEntries);
+    const U64 hashSalt = ms->hashSalt;
     U32 nbAttempts = 1U << cappedSearchLog;
     size_t ml=4-1;
+    U32 hash;
 
     /* DMS/DDS variables that may be referenced laster */
     const ZSTD_matchState_t* const dms = ms->dictMatchState;
@@ -1188,7 +1192,7 @@ size_t ZSTD_RowFindBestMatch(
     if (dictMode == ZSTD_dictMatchState) {
         /* Prefetch DMS rows */
         U32* const dmsHashTable = dms->hashTable;
-        U16* const dmsTagTable = dms->tagTable;
+        BYTE* const dmsTagTable = dms->tagTable;
         U32 const dmsHash = (U32)ZSTD_hashPtr(ip, dms->rowHashLog + ZSTD_ROW_HASH_TAG_BITS, mls);
         U32 const dmsRelRow = (dmsHash >> ZSTD_ROW_HASH_TAG_BITS) << rowLog;
         dmsTag = dmsHash & ZSTD_ROW_HASH_TAG_MASK;
@@ -1198,9 +1202,19 @@ size_t ZSTD_RowFindBestMatch(
     }
 
     /* Update the hashTable and tagTable up to (but not including) ip */
-    ZSTD_row_update_internal(ms, ip, mls, rowLog, rowMask, 1 /* useCache */);
+    if (!ms->lazySkipping) {
+        ZSTD_row_update_internal(ms, ip, mls, rowLog, rowMask, 1 /* useCache */);
+        hash = ZSTD_row_nextCachedHash(hashCache, hashTable, tagTable, base, curr, hashLog, rowLog, mls, hashSalt);
+    } else {
+        /* Stop inserting every position when in the lazy skipping mode.
+         * The hash cache is also not kept up to date in this mode.
+         */
+        hash = (U32)ZSTD_hashPtrSalted(ip, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls, hashSalt);
+        ms->nextToUpdate = curr;
+    }
+    ms->hashSaltEntropy += hash; /* collect salt entropy */
+
     {   /* Get the hash for ip, compute the appropriate row */
-        U32 const hash = ZSTD_row_nextCachedHash(hashCache, hashTable, tagTable, base, curr, hashLog, rowLog, mls);
         U32 const relRow = (hash >> ZSTD_ROW_HASH_TAG_BITS) << rowLog;
         U32 const tag = hash & ZSTD_ROW_HASH_TAG_MASK;
         U32* const row = hashTable + relRow;
@@ -1212,9 +1226,10 @@ size_t ZSTD_RowFindBestMatch(
         ZSTD_VecMask matches = ZSTD_row_getMatchMask(tagRow, (BYTE)tag, headGrouped, rowEntries);
 
         /* Cycle through the matches and prefetch */
-        for (; (matches > 0) && (nbAttempts > 0); --nbAttempts, matches &= (matches - 1)) {
+        for (; (matches > 0) && (nbAttempts > 0); matches &= (matches - 1)) {
             U32 const matchPos = ((headGrouped + ZSTD_VecMask_next(matches)) / groupWidth) & rowMask;
             U32 const matchIndex = row[matchPos];
+            if(matchPos == 0) continue;
             assert(numMatches < rowEntries);
             if (matchIndex < lowLimit)
                 break;
@@ -1224,13 +1239,14 @@ size_t ZSTD_RowFindBestMatch(
                 PREFETCH_L1(dictBase + matchIndex);
             }
             matchBuffer[numMatches++] = matchIndex;
+            --nbAttempts;
         }
 
         /* Speed opt: insert current byte into hashtable too. This allows us to avoid one iteration of the loop
            in ZSTD_row_update_internal() at the next search. */
         {
             U32 const pos = ZSTD_row_nextIndex(tagRow, rowMask);
-            tagRow[pos + ZSTD_ROW_HASH_TAG_OFFSET] = (BYTE)tag;
+            tagRow[pos] = (BYTE)tag;
             row[pos] = ms->nextToUpdate++;
         }
 
@@ -1281,13 +1297,15 @@ size_t ZSTD_RowFindBestMatch(
             size_t currMatch = 0;
             ZSTD_VecMask matches = ZSTD_row_getMatchMask(dmsTagRow, (BYTE)dmsTag, headGrouped, rowEntries);
 
-            for (; (matches > 0) && (nbAttempts > 0); --nbAttempts, matches &= (matches - 1)) {
+            for (; (matches > 0) && (nbAttempts > 0); matches &= (matches - 1)) {
                 U32 const matchPos = ((headGrouped + ZSTD_VecMask_next(matches)) / groupWidth) & rowMask;
                 U32 const matchIndex = dmsRow[matchPos];
+                if(matchPos == 0) continue;
                 if (matchIndex < dmsLowestIndex)
                     break;
                 PREFETCH_L1(dmsBase + matchIndex);
                 matchBuffer[numMatches++] = matchIndex;
+                --nbAttempts;
             }
 
             /* Return the longest match */
@@ -1493,8 +1511,9 @@ FORCE_INLINE_TEMPLATE size_t ZSTD_searchMax(
 *  Common parser - lazy strategy
 *********************************/
 
-FORCE_INLINE_TEMPLATE size_t
-ZSTD_compressBlock_lazy_generic(
+FORCE_INLINE_TEMPLATE
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+size_t ZSTD_compressBlock_lazy_generic(
                         ZSTD_matchState_t* ms, seqStore_t* seqStore,
                         U32 rep[ZSTD_REP_NUM],
                         const void* src, size_t srcSize,
@@ -1544,10 +1563,11 @@ ZSTD_compressBlock_lazy_generic(
         assert(offset_2 <= dictAndPrefixLength);
     }
 
+    /* Reset the lazy skipping state */
+    ms->lazySkipping = 0;
+
     if (searchMethod == search_rowHash) {
-        ZSTD_row_fillHashCache(ms, base, rowLog,
-                            MIN(ms->cParams.minMatch, 6 /* mls caps out at 6 */),
-                            ms->nextToUpdate, ilimit);
+        ZSTD_row_fillHashCache(ms, base, rowLog, mls, ms->nextToUpdate, ilimit);
     }
 
     /* Match Loop */
@@ -1591,7 +1611,16 @@ ZSTD_compressBlock_lazy_generic(
         }
 
         if (matchLength < 4) {
-            ip += ((ip-anchor) >> kSearchStrength) + 1;   /* jump faster over incompressible sections */
+            size_t const step = ((size_t)(ip-anchor) >> kSearchStrength) + 1;   /* jump faster over incompressible sections */;
+            ip += step;
+            /* Enter the lazy skipping mode once we are skipping more than 8 bytes at a time.
+             * In this mode we stop inserting every position into our tables, and only insert
+             * positions that we search, which is one in step positions.
+             * The exact cutoff is flexible, I've just chosen a number that is reasonably high,
+             * so we minimize the compression ratio loss in "normal" scenarios. This mode gets
+             * triggered once we've gone 2KB without finding any matches.
+             */
+            ms->lazySkipping = step > kLazySkippingStep;
             continue;
         }
 
@@ -1695,6 +1724,13 @@ _storeSequence:
             ZSTD_storeSeq(seqStore, litLength, anchor, iend, (U32)offBase, matchLength);
             anchor = ip = start + matchLength;
         }
+        if (ms->lazySkipping) {
+            /* We've found a match, disable lazy skipping mode, and refill the hash cache. */
+            if (searchMethod == search_rowHash) {
+                ZSTD_row_fillHashCache(ms, base, rowLog, mls, ms->nextToUpdate, ilimit);
+            }
+            ms->lazySkipping = 0;
+        }
 
         /* check immediate repcode */
         if (isDxS) {
@@ -1741,29 +1777,10 @@ _storeSequence:
     /* Return the last literals size */
     return (size_t)(iend - anchor);
 }
+#endif /* build exclusions */
 
 
-size_t ZSTD_compressBlock_btlazy2(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize)
-{
-    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_binaryTree, 2, ZSTD_noDict);
-}
-
-size_t ZSTD_compressBlock_lazy2(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize)
-{
-    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2, ZSTD_noDict);
-}
-
-size_t ZSTD_compressBlock_lazy(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize)
-{
-    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1, ZSTD_noDict);
-}
-
+#ifndef ZSTD_EXCLUDE_GREEDY_BLOCK_COMPRESSOR
 size_t ZSTD_compressBlock_greedy(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize)
@@ -1771,27 +1788,6 @@ size_t ZSTD_compressBlock_greedy(
     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0, ZSTD_noDict);
 }
 
-size_t ZSTD_compressBlock_btlazy2_dictMatchState(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize)
-{
-    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_binaryTree, 2, ZSTD_dictMatchState);
-}
-
-size_t ZSTD_compressBlock_lazy2_dictMatchState(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize)
-{
-    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2, ZSTD_dictMatchState);
-}
-
-size_t ZSTD_compressBlock_lazy_dictMatchState(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize)
-{
-    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1, ZSTD_dictMatchState);
-}
-
 size_t ZSTD_compressBlock_greedy_dictMatchState(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize)
@@ -1799,21 +1795,6 @@ size_t ZSTD_compressBlock_greedy_dictMatchState(
     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0, ZSTD_dictMatchState);
 }
 
-
-size_t ZSTD_compressBlock_lazy2_dedicatedDictSearch(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize)
-{
-    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2, ZSTD_dedicatedDictSearch);
-}
-
-size_t ZSTD_compressBlock_lazy_dedicatedDictSearch(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize)
-{
-    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1, ZSTD_dedicatedDictSearch);
-}
-
 size_t ZSTD_compressBlock_greedy_dedicatedDictSearch(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize)
@@ -1821,21 +1802,6 @@ size_t ZSTD_compressBlock_greedy_dedicatedDictSearch(
     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0, ZSTD_dedicatedDictSearch);
 }
 
-/* Row-based matchfinder */
-size_t ZSTD_compressBlock_lazy2_row(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize)
-{
-    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 2, ZSTD_noDict);
-}
-
-size_t ZSTD_compressBlock_lazy_row(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize)
-{
-    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 1, ZSTD_noDict);
-}
-
 size_t ZSTD_compressBlock_greedy_row(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize)
@@ -1843,11 +1809,48 @@ size_t ZSTD_compressBlock_greedy_row(
     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 0, ZSTD_noDict);
 }
 
-size_t ZSTD_compressBlock_lazy2_dictMatchState_row(
+size_t ZSTD_compressBlock_greedy_dictMatchState_row(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize)
 {
-    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 2, ZSTD_dictMatchState);
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 0, ZSTD_dictMatchState);
+}
+
+size_t ZSTD_compressBlock_greedy_dedicatedDictSearch_row(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 0, ZSTD_dedicatedDictSearch);
+}
+#endif
+
+#ifndef ZSTD_EXCLUDE_LAZY_BLOCK_COMPRESSOR
+size_t ZSTD_compressBlock_lazy(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1, ZSTD_noDict);
+}
+
+size_t ZSTD_compressBlock_lazy_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1, ZSTD_dictMatchState);
+}
+
+size_t ZSTD_compressBlock_lazy_dedicatedDictSearch(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1, ZSTD_dedicatedDictSearch);
+}
+
+size_t ZSTD_compressBlock_lazy_row(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 1, ZSTD_noDict);
 }
 
 size_t ZSTD_compressBlock_lazy_dictMatchState_row(
@@ -1857,13 +1860,49 @@ size_t ZSTD_compressBlock_lazy_dictMatchState_row(
     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 1, ZSTD_dictMatchState);
 }
 
-size_t ZSTD_compressBlock_greedy_dictMatchState_row(
+size_t ZSTD_compressBlock_lazy_dedicatedDictSearch_row(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize)
 {
-    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 0, ZSTD_dictMatchState);
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 1, ZSTD_dedicatedDictSearch);
+}
+#endif
+
+#ifndef ZSTD_EXCLUDE_LAZY2_BLOCK_COMPRESSOR
+size_t ZSTD_compressBlock_lazy2(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2, ZSTD_noDict);
 }
 
+size_t ZSTD_compressBlock_lazy2_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2, ZSTD_dictMatchState);
+}
+
+size_t ZSTD_compressBlock_lazy2_dedicatedDictSearch(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2, ZSTD_dedicatedDictSearch);
+}
+
+size_t ZSTD_compressBlock_lazy2_row(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 2, ZSTD_noDict);
+}
+
+size_t ZSTD_compressBlock_lazy2_dictMatchState_row(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 2, ZSTD_dictMatchState);
+}
 
 size_t ZSTD_compressBlock_lazy2_dedicatedDictSearch_row(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
@@ -1871,22 +1910,30 @@ size_t ZSTD_compressBlock_lazy2_dedicatedDictSearch_row(
 {
     return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 2, ZSTD_dedicatedDictSearch);
 }
+#endif
 
-size_t ZSTD_compressBlock_lazy_dedicatedDictSearch_row(
+#ifndef ZSTD_EXCLUDE_BTLAZY2_BLOCK_COMPRESSOR
+size_t ZSTD_compressBlock_btlazy2(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize)
 {
-    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 1, ZSTD_dedicatedDictSearch);
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_binaryTree, 2, ZSTD_noDict);
 }
 
-size_t ZSTD_compressBlock_greedy_dedicatedDictSearch_row(
+size_t ZSTD_compressBlock_btlazy2_dictMatchState(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize)
 {
-    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 0, ZSTD_dedicatedDictSearch);
+    return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_binaryTree, 2, ZSTD_dictMatchState);
 }
+#endif
 
+#if !defined(ZSTD_EXCLUDE_GREEDY_BLOCK_COMPRESSOR) \
+ || !defined(ZSTD_EXCLUDE_LAZY_BLOCK_COMPRESSOR) \
+ || !defined(ZSTD_EXCLUDE_LAZY2_BLOCK_COMPRESSOR) \
+ || !defined(ZSTD_EXCLUDE_BTLAZY2_BLOCK_COMPRESSOR)
 FORCE_INLINE_TEMPLATE
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
 size_t ZSTD_compressBlock_lazy_extDict_generic(
                         ZSTD_matchState_t* ms, seqStore_t* seqStore,
                         U32 rep[ZSTD_REP_NUM],
@@ -1912,12 +1959,13 @@ size_t ZSTD_compressBlock_lazy_extDict_generic(
 
     DEBUGLOG(5, "ZSTD_compressBlock_lazy_extDict_generic (searchFunc=%u)", (U32)searchMethod);
 
+    /* Reset the lazy skipping state */
+    ms->lazySkipping = 0;
+
     /* init */
     ip += (ip == prefixStart);
     if (searchMethod == search_rowHash) {
-        ZSTD_row_fillHashCache(ms, base, rowLog,
-                               MIN(ms->cParams.minMatch, 6 /* mls caps out at 6 */),
-                               ms->nextToUpdate, ilimit);
+        ZSTD_row_fillHashCache(ms, base, rowLog, mls, ms->nextToUpdate, ilimit);
     }
 
     /* Match Loop */
@@ -1955,7 +2003,16 @@ size_t ZSTD_compressBlock_lazy_extDict_generic(
         }
 
         if (matchLength < 4) {
-            ip += ((ip-anchor) >> kSearchStrength) + 1;   /* jump faster over incompressible sections */
+            size_t const step = ((size_t)(ip-anchor) >> kSearchStrength);
+            ip += step + 1;   /* jump faster over incompressible sections */
+            /* Enter the lazy skipping mode once we are skipping more than 8 bytes at a time.
+             * In this mode we stop inserting every position into our tables, and only insert
+             * positions that we search, which is one in step positions.
+             * The exact cutoff is flexible, I've just chosen a number that is reasonably high,
+             * so we minimize the compression ratio loss in "normal" scenarios. This mode gets
+             * triggered once we've gone 2KB without finding any matches.
+             */
+            ms->lazySkipping = step > kLazySkippingStep;
             continue;
         }
 
@@ -2041,6 +2098,13 @@ _storeSequence:
             ZSTD_storeSeq(seqStore, litLength, anchor, iend, (U32)offBase, matchLength);
             anchor = ip = start + matchLength;
         }
+        if (ms->lazySkipping) {
+            /* We've found a match, disable lazy skipping mode, and refill the hash cache. */
+            if (searchMethod == search_rowHash) {
+                ZSTD_row_fillHashCache(ms, base, rowLog, mls, ms->nextToUpdate, ilimit);
+            }
+            ms->lazySkipping = 0;
+        }
 
         /* check immediate repcode */
         while (ip <= ilimit) {
@@ -2071,8 +2135,9 @@ _storeSequence:
     /* Return the last literals size */
     return (size_t)(iend - anchor);
 }
+#endif /* build exclusions */
 
-
+#ifndef ZSTD_EXCLUDE_GREEDY_BLOCK_COMPRESSOR
 size_t ZSTD_compressBlock_greedy_extDict(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize)
@@ -2080,6 +2145,15 @@ size_t ZSTD_compressBlock_greedy_extDict(
     return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0);
 }
 
+size_t ZSTD_compressBlock_greedy_extDict_row(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 0);
+}
+#endif
+
+#ifndef ZSTD_EXCLUDE_LAZY_BLOCK_COMPRESSOR
 size_t ZSTD_compressBlock_lazy_extDict(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize)
@@ -2088,29 +2162,6 @@ size_t ZSTD_compressBlock_lazy_extDict(
     return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1);
 }
 
-size_t ZSTD_compressBlock_lazy2_extDict(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize)
-
-{
-    return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2);
-}
-
-size_t ZSTD_compressBlock_btlazy2_extDict(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize)
-
-{
-    return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_binaryTree, 2);
-}
-
-size_t ZSTD_compressBlock_greedy_extDict_row(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize)
-{
-    return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 0);
-}
-
 size_t ZSTD_compressBlock_lazy_extDict_row(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize)
@@ -2118,6 +2169,16 @@ size_t ZSTD_compressBlock_lazy_extDict_row(
 {
     return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 1);
 }
+#endif
+
+#ifndef ZSTD_EXCLUDE_LAZY2_BLOCK_COMPRESSOR
+size_t ZSTD_compressBlock_lazy2_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+
+{
+    return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2);
+}
 
 size_t ZSTD_compressBlock_lazy2_extDict_row(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
@@ -2125,3 +2186,14 @@ size_t ZSTD_compressBlock_lazy2_extDict_row(
 {
     return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_rowHash, 2);
 }
+#endif
+
+#ifndef ZSTD_EXCLUDE_BTLAZY2_BLOCK_COMPRESSOR
+size_t ZSTD_compressBlock_btlazy2_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize)
+
+{
+    return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_binaryTree, 2);
+}
+#endif

lib/compress/zstd_lazy.h

@@ -27,98 +27,173 @@ extern "C" {
 
 #define ZSTD_ROW_HASH_TAG_BITS 8        /* nb bits to use for the tag */
 
+#if !defined(ZSTD_EXCLUDE_GREEDY_BLOCK_COMPRESSOR) \
+ || !defined(ZSTD_EXCLUDE_LAZY_BLOCK_COMPRESSOR) \
+ || !defined(ZSTD_EXCLUDE_LAZY2_BLOCK_COMPRESSOR) \
+ || !defined(ZSTD_EXCLUDE_BTLAZY2_BLOCK_COMPRESSOR)
 U32 ZSTD_insertAndFindFirstIndex(ZSTD_matchState_t* ms, const BYTE* ip);
 void ZSTD_row_update(ZSTD_matchState_t* const ms, const BYTE* ip);
 
 void ZSTD_dedicatedDictSearch_lazy_loadDictionary(ZSTD_matchState_t* ms, const BYTE* const ip);
 
 void ZSTD_preserveUnsortedMark (U32* const table, U32 const size, U32 const reducerValue);  /*! used in ZSTD_reduceIndex(). preemptively increase value of ZSTD_DUBT_UNSORTED_MARK */
+#endif
 
-size_t ZSTD_compressBlock_btlazy2(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize);
-size_t ZSTD_compressBlock_lazy2(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize);
-size_t ZSTD_compressBlock_lazy(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize);
+#ifndef ZSTD_EXCLUDE_GREEDY_BLOCK_COMPRESSOR
 size_t ZSTD_compressBlock_greedy(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
-size_t ZSTD_compressBlock_lazy2_row(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize);
-size_t ZSTD_compressBlock_lazy_row(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize);
 size_t ZSTD_compressBlock_greedy_row(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
-
-size_t ZSTD_compressBlock_btlazy2_dictMatchState(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize);
-size_t ZSTD_compressBlock_lazy2_dictMatchState(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize);
-size_t ZSTD_compressBlock_lazy_dictMatchState(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize);
 size_t ZSTD_compressBlock_greedy_dictMatchState(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
-size_t ZSTD_compressBlock_lazy2_dictMatchState_row(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize);
-size_t ZSTD_compressBlock_lazy_dictMatchState_row(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize);
 size_t ZSTD_compressBlock_greedy_dictMatchState_row(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
-
-size_t ZSTD_compressBlock_lazy2_dedicatedDictSearch(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize);
-size_t ZSTD_compressBlock_lazy_dedicatedDictSearch(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize);
 size_t ZSTD_compressBlock_greedy_dedicatedDictSearch(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
-size_t ZSTD_compressBlock_lazy2_dedicatedDictSearch_row(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize);
-size_t ZSTD_compressBlock_lazy_dedicatedDictSearch_row(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize);
 size_t ZSTD_compressBlock_greedy_dedicatedDictSearch_row(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
-
 size_t ZSTD_compressBlock_greedy_extDict(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
-size_t ZSTD_compressBlock_lazy_extDict(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize);
-size_t ZSTD_compressBlock_lazy2_extDict(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize);
 size_t ZSTD_compressBlock_greedy_extDict_row(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
+
+#define ZSTD_COMPRESSBLOCK_GREEDY ZSTD_compressBlock_greedy
+#define ZSTD_COMPRESSBLOCK_GREEDY_ROW ZSTD_compressBlock_greedy_row
+#define ZSTD_COMPRESSBLOCK_GREEDY_DICTMATCHSTATE ZSTD_compressBlock_greedy_dictMatchState
+#define ZSTD_COMPRESSBLOCK_GREEDY_DICTMATCHSTATE_ROW ZSTD_compressBlock_greedy_dictMatchState_row
+#define ZSTD_COMPRESSBLOCK_GREEDY_DEDICATEDDICTSEARCH ZSTD_compressBlock_greedy_dedicatedDictSearch
+#define ZSTD_COMPRESSBLOCK_GREEDY_DEDICATEDDICTSEARCH_ROW ZSTD_compressBlock_greedy_dedicatedDictSearch_row
+#define ZSTD_COMPRESSBLOCK_GREEDY_EXTDICT ZSTD_compressBlock_greedy_extDict
+#define ZSTD_COMPRESSBLOCK_GREEDY_EXTDICT_ROW ZSTD_compressBlock_greedy_extDict_row
+#else
+#define ZSTD_COMPRESSBLOCK_GREEDY NULL
+#define ZSTD_COMPRESSBLOCK_GREEDY_ROW NULL
+#define ZSTD_COMPRESSBLOCK_GREEDY_DICTMATCHSTATE NULL
+#define ZSTD_COMPRESSBLOCK_GREEDY_DICTMATCHSTATE_ROW NULL
+#define ZSTD_COMPRESSBLOCK_GREEDY_DEDICATEDDICTSEARCH NULL
+#define ZSTD_COMPRESSBLOCK_GREEDY_DEDICATEDDICTSEARCH_ROW NULL
+#define ZSTD_COMPRESSBLOCK_GREEDY_EXTDICT NULL
+#define ZSTD_COMPRESSBLOCK_GREEDY_EXTDICT_ROW NULL
+#endif
+
+#ifndef ZSTD_EXCLUDE_LAZY_BLOCK_COMPRESSOR
+size_t ZSTD_compressBlock_lazy(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy_row(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy_dictMatchState_row(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy_dedicatedDictSearch(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy_dedicatedDictSearch_row(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
 size_t ZSTD_compressBlock_lazy_extDict_row(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
+
+#define ZSTD_COMPRESSBLOCK_LAZY ZSTD_compressBlock_lazy
+#define ZSTD_COMPRESSBLOCK_LAZY_ROW ZSTD_compressBlock_lazy_row
+#define ZSTD_COMPRESSBLOCK_LAZY_DICTMATCHSTATE ZSTD_compressBlock_lazy_dictMatchState
+#define ZSTD_COMPRESSBLOCK_LAZY_DICTMATCHSTATE_ROW ZSTD_compressBlock_lazy_dictMatchState_row
+#define ZSTD_COMPRESSBLOCK_LAZY_DEDICATEDDICTSEARCH ZSTD_compressBlock_lazy_dedicatedDictSearch
+#define ZSTD_COMPRESSBLOCK_LAZY_DEDICATEDDICTSEARCH_ROW ZSTD_compressBlock_lazy_dedicatedDictSearch_row
+#define ZSTD_COMPRESSBLOCK_LAZY_EXTDICT ZSTD_compressBlock_lazy_extDict
+#define ZSTD_COMPRESSBLOCK_LAZY_EXTDICT_ROW ZSTD_compressBlock_lazy_extDict_row
+#else
+#define ZSTD_COMPRESSBLOCK_LAZY NULL
+#define ZSTD_COMPRESSBLOCK_LAZY_ROW NULL
+#define ZSTD_COMPRESSBLOCK_LAZY_DICTMATCHSTATE NULL
+#define ZSTD_COMPRESSBLOCK_LAZY_DICTMATCHSTATE_ROW NULL
+#define ZSTD_COMPRESSBLOCK_LAZY_DEDICATEDDICTSEARCH NULL
+#define ZSTD_COMPRESSBLOCK_LAZY_DEDICATEDDICTSEARCH_ROW NULL
+#define ZSTD_COMPRESSBLOCK_LAZY_EXTDICT NULL
+#define ZSTD_COMPRESSBLOCK_LAZY_EXTDICT_ROW NULL
+#endif
+
+#ifndef ZSTD_EXCLUDE_LAZY2_BLOCK_COMPRESSOR
+size_t ZSTD_compressBlock_lazy2(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy2_row(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy2_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy2_dictMatchState_row(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy2_dedicatedDictSearch(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy2_dedicatedDictSearch_row(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_lazy2_extDict(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
 size_t ZSTD_compressBlock_lazy2_extDict_row(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
+
+#define ZSTD_COMPRESSBLOCK_LAZY2 ZSTD_compressBlock_lazy2
+#define ZSTD_COMPRESSBLOCK_LAZY2_ROW ZSTD_compressBlock_lazy2_row
+#define ZSTD_COMPRESSBLOCK_LAZY2_DICTMATCHSTATE ZSTD_compressBlock_lazy2_dictMatchState
+#define ZSTD_COMPRESSBLOCK_LAZY2_DICTMATCHSTATE_ROW ZSTD_compressBlock_lazy2_dictMatchState_row
+#define ZSTD_COMPRESSBLOCK_LAZY2_DEDICATEDDICTSEARCH ZSTD_compressBlock_lazy2_dedicatedDictSearch
+#define ZSTD_COMPRESSBLOCK_LAZY2_DEDICATEDDICTSEARCH_ROW ZSTD_compressBlock_lazy2_dedicatedDictSearch_row
+#define ZSTD_COMPRESSBLOCK_LAZY2_EXTDICT ZSTD_compressBlock_lazy2_extDict
+#define ZSTD_COMPRESSBLOCK_LAZY2_EXTDICT_ROW ZSTD_compressBlock_lazy2_extDict_row
+#else
+#define ZSTD_COMPRESSBLOCK_LAZY2 NULL
+#define ZSTD_COMPRESSBLOCK_LAZY2_ROW NULL
+#define ZSTD_COMPRESSBLOCK_LAZY2_DICTMATCHSTATE NULL
+#define ZSTD_COMPRESSBLOCK_LAZY2_DICTMATCHSTATE_ROW NULL
+#define ZSTD_COMPRESSBLOCK_LAZY2_DEDICATEDDICTSEARCH NULL
+#define ZSTD_COMPRESSBLOCK_LAZY2_DEDICATEDDICTSEARCH_ROW NULL
+#define ZSTD_COMPRESSBLOCK_LAZY2_EXTDICT NULL
+#define ZSTD_COMPRESSBLOCK_LAZY2_EXTDICT_ROW NULL
+#endif
+
+#ifndef ZSTD_EXCLUDE_BTLAZY2_BLOCK_COMPRESSOR
+size_t ZSTD_compressBlock_btlazy2(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_btlazy2_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
 size_t ZSTD_compressBlock_btlazy2_extDict(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
 
+#define ZSTD_COMPRESSBLOCK_BTLAZY2 ZSTD_compressBlock_btlazy2
+#define ZSTD_COMPRESSBLOCK_BTLAZY2_DICTMATCHSTATE ZSTD_compressBlock_btlazy2_dictMatchState
+#define ZSTD_COMPRESSBLOCK_BTLAZY2_EXTDICT ZSTD_compressBlock_btlazy2_extDict
+#else
+#define ZSTD_COMPRESSBLOCK_BTLAZY2 NULL
+#define ZSTD_COMPRESSBLOCK_BTLAZY2_DICTMATCHSTATE NULL
+#define ZSTD_COMPRESSBLOCK_BTLAZY2_EXTDICT NULL
+#endif
+
 
 #if defined (__cplusplus)
 }

lib/compress/zstd_ldm.c

@@ -246,7 +246,11 @@ static size_t ZSTD_ldm_fillFastTables(ZSTD_matchState_t* ms,
         break;
 
     case ZSTD_dfast:
+#ifndef ZSTD_EXCLUDE_DFAST_BLOCK_COMPRESSOR
         ZSTD_fillDoubleHashTable(ms, iend, ZSTD_dtlm_fast, ZSTD_tfp_forCCtx);
+#else
+        assert(0); /* shouldn't be called: cparams should've been adjusted. */
+#endif
         break;
 
     case ZSTD_greedy:
@@ -318,7 +322,9 @@ static void ZSTD_ldm_limitTableUpdate(ZSTD_matchState_t* ms, const BYTE* anchor)
     }
 }
 
-static size_t ZSTD_ldm_generateSequences_internal(
+static
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+size_t ZSTD_ldm_generateSequences_internal(
         ldmState_t* ldmState, rawSeqStore_t* rawSeqStore,
         ldmParams_t const* params, void const* src, size_t srcSize)
 {
@@ -689,7 +695,6 @@ size_t ZSTD_ldm_blockCompress(rawSeqStore_t* rawSeqStore,
         /* maybeSplitSequence updates rawSeqStore->pos */
         rawSeq const sequence = maybeSplitSequence(rawSeqStore,
                                                    (U32)(iend - ip), minMatch);
-        int i;
         /* End signal */
         if (sequence.offset == 0)
             break;
@@ -702,6 +707,7 @@ size_t ZSTD_ldm_blockCompress(rawSeqStore_t* rawSeqStore,
         /* Run the block compressor */
         DEBUGLOG(5, "pos %u : calling block compressor on segment of size %u", (unsigned)(ip-istart), sequence.litLength);
         {
+            int i;
             size_t const newLitLength =
                 blockCompressor(ms, seqStore, rep, ip, sequence.litLength);
             ip += sequence.litLength;

lib/compress/zstd_opt.c

@@ -12,6 +12,9 @@
 #include "hist.h"
 #include "zstd_opt.h"
 
+#if !defined(ZSTD_EXCLUDE_BTLAZY2_BLOCK_COMPRESSOR) \
+ || !defined(ZSTD_EXCLUDE_BTOPT_BLOCK_COMPRESSOR) \
+ || !defined(ZSTD_EXCLUDE_BTULTRA_BLOCK_COMPRESSOR)
 
 #define ZSTD_LITFREQ_ADD    2   /* scaling factor for litFreq, so that frequencies adapt faster to new stats */
 #define ZSTD_MAX_PRICE     (1<<30)
@@ -264,6 +267,7 @@ static U32 ZSTD_rawLiteralsCost(const BYTE* const literals, U32 const litLength,
                                 const optState_t* const optPtr,
                                 int optLevel)
 {
+    DEBUGLOG(8, "ZSTD_rawLiteralsCost (%u literals)", litLength);
     if (litLength == 0) return 0;
 
     if (!ZSTD_compressedLiterals(optPtr))
@@ -402,9 +406,11 @@ MEM_STATIC U32 ZSTD_readMINMATCH(const void* memPtr, U32 length)
 
 /* Update hashTable3 up to ip (excluded)
    Assumption : always within prefix (i.e. not within extDict) */
-static U32 ZSTD_insertAndFindFirstIndexHash3 (const ZSTD_matchState_t* ms,
-                                              U32* nextToUpdate3,
-                                              const BYTE* const ip)
+static
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+U32 ZSTD_insertAndFindFirstIndexHash3 (const ZSTD_matchState_t* ms,
+                                       U32* nextToUpdate3,
+                                       const BYTE* const ip)
 {
     U32* const hashTable3 = ms->hashTable3;
     U32 const hashLog3 = ms->hashLog3;
@@ -431,7 +437,9 @@ static U32 ZSTD_insertAndFindFirstIndexHash3 (const ZSTD_matchState_t* ms,
  * @param ip assumed <= iend-8 .
  * @param target The target of ZSTD_updateTree_internal() - we are filling to this position
  * @return : nb of positions added */
-static U32 ZSTD_insertBt1(
+static
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+U32 ZSTD_insertBt1(
                 const ZSTD_matchState_t* ms,
                 const BYTE* const ip, const BYTE* const iend,
                 U32 const target,
@@ -550,6 +558,7 @@ static U32 ZSTD_insertBt1(
 }
 
 FORCE_INLINE_TEMPLATE
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
 void ZSTD_updateTree_internal(
                 ZSTD_matchState_t* ms,
                 const BYTE* const ip, const BYTE* const iend,
@@ -558,7 +567,7 @@ void ZSTD_updateTree_internal(
     const BYTE* const base = ms->window.base;
     U32 const target = (U32)(ip - base);
     U32 idx = ms->nextToUpdate;
-    DEBUGLOG(6, "ZSTD_updateTree_internal, from %u to %u  (dictMode:%u)",
+    DEBUGLOG(7, "ZSTD_updateTree_internal, from %u to %u  (dictMode:%u)",
                 idx, target, dictMode);
 
     while(idx < target) {
@@ -575,7 +584,9 @@ void ZSTD_updateTree(ZSTD_matchState_t* ms, const BYTE* ip, const BYTE* iend) {
     ZSTD_updateTree_internal(ms, ip, iend, ms->cParams.minMatch, ZSTD_noDict);
 }
 
-FORCE_INLINE_TEMPLATE U32
+FORCE_INLINE_TEMPLATE
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+U32
 ZSTD_insertBtAndGetAllMatches (
                 ZSTD_match_t* matches,  /* store result (found matches) in this table (presumed large enough) */
                 ZSTD_matchState_t* ms,
@@ -816,7 +827,9 @@ typedef U32 (*ZSTD_getAllMatchesFn)(
     U32 const ll0,
     U32 const lengthToBeat);
 
-FORCE_INLINE_TEMPLATE U32 ZSTD_btGetAllMatches_internal(
+FORCE_INLINE_TEMPLATE
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+U32 ZSTD_btGetAllMatches_internal(
         ZSTD_match_t* matches,
         ZSTD_matchState_t* ms,
         U32* nextToUpdate3,
@@ -1035,11 +1048,6 @@ ZSTD_optLdm_processMatchCandidate(ZSTD_optLdm_t* optLdm,
 *  Optimal parser
 *********************************/
 
-static U32 ZSTD_totalLen(ZSTD_optimal_t sol)
-{
-    return sol.litlen + sol.mlen;
-}
-
 #if 0 /* debug */
 
 static void
@@ -1057,7 +1065,13 @@ listStats(const U32* table, int lastEltID)
 
 #endif
 
-FORCE_INLINE_TEMPLATE size_t
+#define LIT_PRICE(_p) (int)ZSTD_rawLiteralsCost(_p, 1, optStatePtr, optLevel)
+#define LL_PRICE(_l) (int)ZSTD_litLengthPrice(_l, optStatePtr, optLevel)
+#define LL_INCPRICE(_l) (LL_PRICE(_l) - LL_PRICE(_l-1))
+
+FORCE_INLINE_TEMPLATE
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+size_t
 ZSTD_compressBlock_opt_generic(ZSTD_matchState_t* ms,
                                seqStore_t* seqStore,
                                U32 rep[ZSTD_REP_NUM],
@@ -1083,9 +1097,11 @@ ZSTD_compressBlock_opt_generic(ZSTD_matchState_t* ms,
 
     ZSTD_optimal_t* const opt = optStatePtr->priceTable;
     ZSTD_match_t* const matches = optStatePtr->matchTable;
-    ZSTD_optimal_t lastSequence;
+    ZSTD_optimal_t lastStretch;
     ZSTD_optLdm_t optLdm;
 
+    ZSTD_memset(&lastStretch, 0, sizeof(ZSTD_optimal_t));
+
     optLdm.seqStore = ms->ldmSeqStore ? *ms->ldmSeqStore : kNullRawSeqStore;
     optLdm.endPosInBlock = optLdm.startPosInBlock = optLdm.offset = 0;
     ZSTD_opt_getNextMatchAndUpdateSeqStore(&optLdm, (U32)(ip-istart), (U32)(iend-ip));
@@ -1106,19 +1122,31 @@ ZSTD_compressBlock_opt_generic(ZSTD_matchState_t* ms,
             U32 const ll0 = !litlen;
             U32 nbMatches = getAllMatches(matches, ms, &nextToUpdate3, ip, iend, rep, ll0, minMatch);
             ZSTD_optLdm_processMatchCandidate(&optLdm, matches, &nbMatches,
-                                              (U32)(ip-istart), (U32)(iend - ip));
-            if (!nbMatches) { ip++; continue; }
+                                              (U32)(ip-istart), (U32)(iend-ip));
+            if (!nbMatches) {
+                DEBUGLOG(8, "no match found at cPos %u", (unsigned)(ip-istart));
+                ip++;
+                continue;
+            }
+
+            /* Match found: let's store this solution, and eventually find more candidates.
+             * During this forward pass, @opt is used to store stretches,
+             * defined as "a match followed by N literals".
+             * Note how this is different from a Sequence, which is "N literals followed by a match".
+             * Storing stretches allows us to store different match predecessors
+             * for each literal position part of a literals run. */
 
             /* initialize opt[0] */
-            { U32 i ; for (i=0; i<ZSTD_REP_NUM; i++) opt[0].rep[i] = rep[i]; }
-            opt[0].mlen = 0;  /* means is_a_literal */
+            opt[0].mlen = 0;  /* there are only literals so far */
             opt[0].litlen = litlen;
-            /* We don't need to include the actual price of the literals because
-             * it is static for the duration of the forward pass, and is included
-             * in every price. We include the literal length to avoid negative
-             * prices when we subtract the previous literal length.
+            /* No need to include the actual price of the literals before the first match
+             * because it is static for the duration of the forward pass, and is included
+             * in every subsequent price. But, we include the literal length because
+             * the cost variation of litlen depends on the value of litlen.
              */
-            opt[0].price = (int)ZSTD_litLengthPrice(litlen, optStatePtr, optLevel);
+            opt[0].price = LL_PRICE(litlen);
+            ZSTD_STATIC_ASSERT(sizeof(opt[0].rep[0]) == sizeof(rep[0]));
+            ZSTD_memcpy(&opt[0].rep, rep, sizeof(opt[0].rep));
 
             /* large match -> immediate encoding */
             {   U32 const maxML = matches[nbMatches-1].len;
@@ -1127,82 +1155,106 @@ ZSTD_compressBlock_opt_generic(ZSTD_matchState_t* ms,
                             nbMatches, maxML, maxOffBase, (U32)(ip-prefixStart));
 
                 if (maxML > sufficient_len) {
-                    lastSequence.litlen = litlen;
-                    lastSequence.mlen = maxML;
-                    lastSequence.off = maxOffBase;
-                    DEBUGLOG(6, "large match (%u>%u), immediate encoding",
+                    lastStretch.litlen = 0;
+                    lastStretch.mlen = maxML;
+                    lastStretch.off = maxOffBase;
+                    DEBUGLOG(6, "large match (%u>%u) => immediate encoding",
                                 maxML, sufficient_len);
                     cur = 0;
-                    last_pos = ZSTD_totalLen(lastSequence);
+                    last_pos = maxML;
                     goto _shortestPath;
             }   }
 
             /* set prices for first matches starting position == 0 */
             assert(opt[0].price >= 0);
-            {   U32 const literalsPrice = (U32)opt[0].price + ZSTD_litLengthPrice(0, optStatePtr, optLevel);
-                U32 pos;
+            {   U32 pos;
                 U32 matchNb;
                 for (pos = 1; pos < minMatch; pos++) {
-                    opt[pos].price = ZSTD_MAX_PRICE;   /* mlen, litlen and price will be fixed during forward scanning */
+                    opt[pos].price = ZSTD_MAX_PRICE;
+                    opt[pos].mlen = 0;
+                    opt[pos].litlen = litlen + pos;
                 }
                 for (matchNb = 0; matchNb < nbMatches; matchNb++) {
                     U32 const offBase = matches[matchNb].off;
                     U32 const end = matches[matchNb].len;
                     for ( ; pos <= end ; pos++ ) {
-                        U32 const matchPrice = ZSTD_getMatchPrice(offBase, pos, optStatePtr, optLevel);
-                        U32 const sequencePrice = literalsPrice + matchPrice;
+                        int const matchPrice = (int)ZSTD_getMatchPrice(offBase, pos, optStatePtr, optLevel);
+                        int const sequencePrice = opt[0].price + matchPrice;
                         DEBUGLOG(7, "rPos:%u => set initial price : %.2f",
-                                    pos, ZSTD_fCost((int)sequencePrice));
+                                    pos, ZSTD_fCost(sequencePrice));
                         opt[pos].mlen = pos;
                         opt[pos].off = offBase;
-                        opt[pos].litlen = litlen;
-                        opt[pos].price = (int)sequencePrice;
-                }   }
+                        opt[pos].litlen = 0; /* end of match */
+                        opt[pos].price = sequencePrice + LL_PRICE(0);
+                    }
+                }
                 last_pos = pos-1;
+                opt[pos].price = ZSTD_MAX_PRICE;
             }
         }
 
         /* check further positions */
         for (cur = 1; cur <= last_pos; cur++) {
             const BYTE* const inr = ip + cur;
-            assert(cur < ZSTD_OPT_NUM);
-            DEBUGLOG(7, "cPos:%zi==rPos:%u", inr-istart, cur)
+            assert(cur <= ZSTD_OPT_NUM);
+            DEBUGLOG(7, "cPos:%zi==rPos:%u", inr-istart, cur);
 
             /* Fix current position with one literal if cheaper */
-            {   U32 const litlen = (opt[cur-1].mlen == 0) ? opt[cur-1].litlen + 1 : 1;
+            {   U32 const litlen = opt[cur-1].litlen + 1;
                 int const price = opt[cur-1].price
-                                + (int)ZSTD_rawLiteralsCost(ip+cur-1, 1, optStatePtr, optLevel)
-                                + (int)ZSTD_litLengthPrice(litlen, optStatePtr, optLevel)
-                                - (int)ZSTD_litLengthPrice(litlen-1, optStatePtr, optLevel);
+                                + LIT_PRICE(ip+cur-1)
+                                + LL_INCPRICE(litlen);
                 assert(price < 1000000000); /* overflow check */
                 if (price <= opt[cur].price) {
+                    ZSTD_optimal_t const prevMatch = opt[cur];
                     DEBUGLOG(7, "cPos:%zi==rPos:%u : better price (%.2f<=%.2f) using literal (ll==%u) (hist:%u,%u,%u)",
                                 inr-istart, cur, ZSTD_fCost(price), ZSTD_fCost(opt[cur].price), litlen,
                                 opt[cur-1].rep[0], opt[cur-1].rep[1], opt[cur-1].rep[2]);
-                    opt[cur].mlen = 0;
-                    opt[cur].off = 0;
+                    opt[cur] = opt[cur-1];
                     opt[cur].litlen = litlen;
                     opt[cur].price = price;
+                    if ( (optLevel >= 1) /* additional check only for higher modes */
+                      && (prevMatch.litlen == 0) /* replace a match */
+                      && (LL_INCPRICE(1) < 0) /* ll1 is cheaper than ll0 */
+                      && LIKELY(ip + cur < iend)
+                    ) {
+                        /* check next position, in case it would be cheaper */
+                        int with1literal = prevMatch.price + LIT_PRICE(ip+cur) + LL_INCPRICE(1);
+                        int withMoreLiterals = price + LIT_PRICE(ip+cur) + LL_INCPRICE(litlen+1);
+                        DEBUGLOG(7, "then at next rPos %u : match+1lit %.2f vs %ulits %.2f",
+                                cur+1, ZSTD_fCost(with1literal), litlen+1, ZSTD_fCost(withMoreLiterals));
+                        if ( (with1literal < withMoreLiterals)
+                          && (with1literal < opt[cur+1].price) ) {
+                            /* update offset history - before it disappears */
+                            U32 const prev = cur - prevMatch.mlen;
+                            repcodes_t const newReps = ZSTD_newRep(opt[prev].rep, prevMatch.off, opt[prev].litlen==0);
+                            assert(cur >= prevMatch.mlen);
+                            DEBUGLOG(7, "==> match+1lit is cheaper (%.2f < %.2f) (hist:%u,%u,%u) !",
+                                        ZSTD_fCost(with1literal), ZSTD_fCost(withMoreLiterals),
+                                        newReps.rep[0], newReps.rep[1], newReps.rep[2] );
+                            opt[cur+1] = prevMatch;  /* mlen & offbase */
+                            ZSTD_memcpy(opt[cur+1].rep, &newReps, sizeof(repcodes_t));
+                            opt[cur+1].litlen = 1;
+                            opt[cur+1].price = with1literal;
+                            if (last_pos < cur+1) last_pos = cur+1;
+                        }
+                    }
                 } else {
-                    DEBUGLOG(7, "cPos:%zi==rPos:%u : literal would cost more (%.2f>%.2f) (hist:%u,%u,%u)",
-                                inr-istart, cur, ZSTD_fCost(price), ZSTD_fCost(opt[cur].price),
-                                opt[cur].rep[0], opt[cur].rep[1], opt[cur].rep[2]);
+                    DEBUGLOG(7, "cPos:%zi==rPos:%u : literal would cost more (%.2f>%.2f)",
+                                inr-istart, cur, ZSTD_fCost(price), ZSTD_fCost(opt[cur].price));
                 }
             }
 
-            /* Set the repcodes of the current position. We must do it here
-             * because we rely on the repcodes of the 2nd to last sequence being
-             * correct to set the next chunks repcodes during the backward
-             * traversal.
+            /* Offset history is not updated during match comparison.
+             * Do it here, now that the match is selected and confirmed.
              */
             ZSTD_STATIC_ASSERT(sizeof(opt[cur].rep) == sizeof(repcodes_t));
             assert(cur >= opt[cur].mlen);
-            if (opt[cur].mlen != 0) {
+            if (opt[cur].litlen == 0) {
+                /* just finished a match => alter offset history */
                 U32 const prev = cur - opt[cur].mlen;
-                repcodes_t const newReps = ZSTD_newRep(opt[prev].rep, opt[cur].off, opt[cur].litlen==0);
+                repcodes_t const newReps = ZSTD_newRep(opt[prev].rep, opt[cur].off, opt[prev].litlen==0);
                 ZSTD_memcpy(opt[cur].rep, &newReps, sizeof(repcodes_t));
-            } else {
-                ZSTD_memcpy(opt[cur].rep, opt[cur - 1].rep, sizeof(repcodes_t));
             }
 
             /* last match must start at a minimum distance of 8 from oend */
@@ -1212,15 +1264,14 @@ ZSTD_compressBlock_opt_generic(ZSTD_matchState_t* ms,
 
             if ( (optLevel==0) /*static_test*/
               && (opt[cur+1].price <= opt[cur].price + (BITCOST_MULTIPLIER/2)) ) {
-                DEBUGLOG(7, "move to next rPos:%u : price is <=", cur+1);
+                DEBUGLOG(7, "skip current position : next rPos(%u) price is cheaper", cur+1);
                 continue;  /* skip unpromising positions; about ~+6% speed, -0.01 ratio */
             }
 
             assert(opt[cur].price >= 0);
-            {   U32 const ll0 = (opt[cur].mlen != 0);
-                U32 const litlen = (opt[cur].mlen == 0) ? opt[cur].litlen : 0;
-                U32 const previousPrice = (U32)opt[cur].price;
-                U32 const basePrice = previousPrice + ZSTD_litLengthPrice(0, optStatePtr, optLevel);
+            {   U32 const ll0 = (opt[cur].litlen == 0);
+                int const previousPrice = opt[cur].price;
+                int const basePrice = previousPrice + LL_PRICE(0);
                 U32 nbMatches = getAllMatches(matches, ms, &nextToUpdate3, inr, iend, opt[cur].rep, ll0, minMatch);
                 U32 matchNb;
 
@@ -1232,18 +1283,17 @@ ZSTD_compressBlock_opt_generic(ZSTD_matchState_t* ms,
                     continue;
                 }
 
-                {   U32 const maxML = matches[nbMatches-1].len;
-                    DEBUGLOG(7, "cPos:%zi==rPos:%u, found %u matches, of maxLength=%u",
-                                inr-istart, cur, nbMatches, maxML);
+                {   U32 const longestML = matches[nbMatches-1].len;
+                    DEBUGLOG(7, "cPos:%zi==rPos:%u, found %u matches, of longest ML=%u",
+                                inr-istart, cur, nbMatches, longestML);
 
-                    if ( (maxML > sufficient_len)
-                      || (cur + maxML >= ZSTD_OPT_NUM) ) {
-                        lastSequence.mlen = maxML;
-                        lastSequence.off = matches[nbMatches-1].off;
-                        lastSequence.litlen = litlen;
-                        cur -= (opt[cur].mlen==0) ? opt[cur].litlen : 0;  /* last sequence is actually only literals, fix cur to last match - note : may underflow, in which case, it's first sequence, and it's okay */
-                        last_pos = cur + ZSTD_totalLen(lastSequence);
-                        if (cur > ZSTD_OPT_NUM) cur = 0;   /* underflow => first match */
+                    if ( (longestML > sufficient_len)
+                      || (cur + longestML >= ZSTD_OPT_NUM)
+                      || (ip + cur + longestML >= iend) ) {
+                        lastStretch.mlen = longestML;
+                        lastStretch.off = matches[nbMatches-1].off;
+                        lastStretch.litlen = 0;
+                        last_pos = cur + longestML;
                         goto _shortestPath;
                 }   }
 
@@ -1255,19 +1305,24 @@ ZSTD_compressBlock_opt_generic(ZSTD_matchState_t* ms,
                     U32 mlen;
 
                     DEBUGLOG(7, "testing match %u => offBase=%4u, mlen=%2u, llen=%2u",
-                                matchNb, matches[matchNb].off, lastML, litlen);
+                                matchNb, matches[matchNb].off, lastML, opt[cur].litlen);
 
                     for (mlen = lastML; mlen >= startML; mlen--) {  /* scan downward */
                         U32 const pos = cur + mlen;
-                        int const price = (int)basePrice + (int)ZSTD_getMatchPrice(offset, mlen, optStatePtr, optLevel);
+                        int const price = basePrice + (int)ZSTD_getMatchPrice(offset, mlen, optStatePtr, optLevel);
 
                         if ((pos > last_pos) || (price < opt[pos].price)) {
                             DEBUGLOG(7, "rPos:%u (ml=%2u) => new better price (%.2f<%.2f)",
                                         pos, mlen, ZSTD_fCost(price), ZSTD_fCost(opt[pos].price));
-                            while (last_pos < pos) { opt[last_pos+1].price = ZSTD_MAX_PRICE; last_pos++; }   /* fill empty positions */
+                            while (last_pos < pos) {
+                                /* fill empty positions, for future comparisons */
+                                last_pos++;
+                                opt[last_pos].price = ZSTD_MAX_PRICE;
+                                opt[last_pos].litlen = !0;  /* just needs to be != 0, to mean "not an end of match" */
+                            }
                             opt[pos].mlen = mlen;
                             opt[pos].off = offset;
-                            opt[pos].litlen = litlen;
+                            opt[pos].litlen = 0;
                             opt[pos].price = price;
                         } else {
                             DEBUGLOG(7, "rPos:%u (ml=%2u) => new price is worse (%.2f>=%.2f)",
@@ -1275,47 +1330,81 @@ ZSTD_compressBlock_opt_generic(ZSTD_matchState_t* ms,
                             if (optLevel==0) break;  /* early update abort; gets ~+10% speed for about -0.01 ratio loss */
                         }
             }   }   }
+            opt[last_pos+1].price = ZSTD_MAX_PRICE;
         }  /* for (cur = 1; cur <= last_pos; cur++) */
 
-        lastSequence = opt[last_pos];
-        cur = last_pos > ZSTD_totalLen(lastSequence) ? last_pos - ZSTD_totalLen(lastSequence) : 0;  /* single sequence, and it starts before `ip` */
-        assert(cur < ZSTD_OPT_NUM);  /* control overflow*/
+        lastStretch = opt[last_pos];
+        assert(cur >= lastStretch.mlen);
+        cur = last_pos - lastStretch.mlen;
 
 _shortestPath:   /* cur, last_pos, best_mlen, best_off have to be set */
         assert(opt[0].mlen == 0);
+        assert(last_pos >= lastStretch.mlen);
+        assert(cur == last_pos - lastStretch.mlen);
 
-        /* Set the next chunk's repcodes based on the repcodes of the beginning
-         * of the last match, and the last sequence. This avoids us having to
-         * update them while traversing the sequences.
-         */
-        if (lastSequence.mlen != 0) {
-            repcodes_t const reps = ZSTD_newRep(opt[cur].rep, lastSequence.off, lastSequence.litlen==0);
-            ZSTD_memcpy(rep, &reps, sizeof(reps));
+        if (lastStretch.mlen==0) {
+            /* no solution : all matches have been converted into literals */
+            assert(lastStretch.litlen == (ip - anchor) + last_pos);
+            ip += last_pos;
+            continue;
+        }
+        assert(lastStretch.off > 0);
+
+        /* Update offset history */
+        if (lastStretch.litlen == 0) {
+            /* finishing on a match : update offset history */
+            repcodes_t const reps = ZSTD_newRep(opt[cur].rep, lastStretch.off, opt[cur].litlen==0);
+            ZSTD_memcpy(rep, &reps, sizeof(repcodes_t));
         } else {
-            ZSTD_memcpy(rep, opt[cur].rep, sizeof(repcodes_t));
+            ZSTD_memcpy(rep, lastStretch.rep, sizeof(repcodes_t));
+            assert(cur >= lastStretch.litlen);
+            cur -= lastStretch.litlen;
         }
 
-        {   U32 const storeEnd = cur + 1;
+        /* Let's write the shortest path solution.
+         * It is stored in @opt in reverse order,
+         * starting from @storeEnd (==cur+2),
+         * effectively partially @opt overwriting.
+         * Content is changed too:
+         * - So far, @opt stored stretches, aka a match followed by literals
+         * - Now, it will store sequences, aka literals followed by a match
+         */
+        {   U32 const storeEnd = cur + 2;
             U32 storeStart = storeEnd;
-            U32 seqPos = cur;
+            U32 stretchPos = cur;
 
             DEBUGLOG(6, "start reverse traversal (last_pos:%u, cur:%u)",
                         last_pos, cur); (void)last_pos;
-            assert(storeEnd < ZSTD_OPT_NUM);
-            DEBUGLOG(6, "last sequence copied into pos=%u (llen=%u,mlen=%u,ofc=%u)",
-                        storeEnd, lastSequence.litlen, lastSequence.mlen, lastSequence.off);
-            opt[storeEnd] = lastSequence;
-            while (seqPos > 0) {
-                U32 const backDist = ZSTD_totalLen(opt[seqPos]);
+            assert(storeEnd < ZSTD_OPT_SIZE);
+            DEBUGLOG(6, "last stretch copied into pos=%u (llen=%u,mlen=%u,ofc=%u)",
+                        storeEnd, lastStretch.litlen, lastStretch.mlen, lastStretch.off);
+            if (lastStretch.litlen > 0) {
+                /* last "sequence" is unfinished: just a bunch of literals */
+                opt[storeEnd].litlen = lastStretch.litlen;
+                opt[storeEnd].mlen = 0;
+                storeStart = storeEnd-1;
+                opt[storeStart] = lastStretch;
+            } {
+                opt[storeEnd] = lastStretch;  /* note: litlen will be fixed */
+                storeStart = storeEnd;
+            }
+            while (1) {
+                ZSTD_optimal_t nextStretch = opt[stretchPos];
+                opt[storeStart].litlen = nextStretch.litlen;
+                DEBUGLOG(6, "selected sequence (llen=%u,mlen=%u,ofc=%u)",
+                            opt[storeStart].litlen, opt[storeStart].mlen, opt[storeStart].off);
+                if (nextStretch.mlen == 0) {
+                    /* reaching beginning of segment */
+                    break;
+                }
                 storeStart--;
-                DEBUGLOG(6, "sequence from rPos=%u copied into pos=%u (llen=%u,mlen=%u,ofc=%u)",
-                            seqPos, storeStart, opt[seqPos].litlen, opt[seqPos].mlen, opt[seqPos].off);
-                opt[storeStart] = opt[seqPos];
-                seqPos = (seqPos > backDist) ? seqPos - backDist : 0;
+                opt[storeStart] = nextStretch; /* note: litlen will be fixed */
+                assert(nextStretch.litlen + nextStretch.mlen <= stretchPos);
+                stretchPos -= nextStretch.litlen + nextStretch.mlen;
             }
 
             /* save sequences */
-            DEBUGLOG(6, "sending selected sequences into seqStore")
+            DEBUGLOG(6, "sending selected sequences into seqStore");
             {   U32 storePos;
                 for (storePos=storeStart; storePos <= storeEnd; storePos++) {
                     U32 const llen = opt[storePos].litlen;
@@ -1337,6 +1426,9 @@ _shortestPath:   /* cur, last_pos, best_mlen, best_off have to be set */
                     anchor += advance;
                     ip = anchor;
             }   }
+            DEBUGLOG(7, "new offset history : %u, %u, %u", rep[0], rep[1], rep[2]);
+
+            /* update all costs */
             ZSTD_setBasePrices(optStatePtr, optLevel);
         }
     }   /* while (ip < ilimit) */
@@ -1344,21 +1436,27 @@ _shortestPath:   /* cur, last_pos, best_mlen, best_off have to be set */
     /* Return the last literals size */
     return (size_t)(iend - anchor);
 }
+#endif /* build exclusions */
 
+#ifndef ZSTD_EXCLUDE_BTOPT_BLOCK_COMPRESSOR
 static size_t ZSTD_compressBlock_opt0(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         const void* src, size_t srcSize, const ZSTD_dictMode_e dictMode)
 {
     return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 0 /* optLevel */, dictMode);
 }
+#endif
 
+#ifndef ZSTD_EXCLUDE_BTULTRA_BLOCK_COMPRESSOR
 static size_t ZSTD_compressBlock_opt2(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         const void* src, size_t srcSize, const ZSTD_dictMode_e dictMode)
 {
     return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 2 /* optLevel */, dictMode);
 }
+#endif
 
+#ifndef ZSTD_EXCLUDE_BTOPT_BLOCK_COMPRESSOR
 size_t ZSTD_compressBlock_btopt(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         const void* src, size_t srcSize)
@@ -1366,20 +1464,23 @@ size_t ZSTD_compressBlock_btopt(
     DEBUGLOG(5, "ZSTD_compressBlock_btopt");
     return ZSTD_compressBlock_opt0(ms, seqStore, rep, src, srcSize, ZSTD_noDict);
 }
+#endif
 
 
 
 
+#ifndef ZSTD_EXCLUDE_BTULTRA_BLOCK_COMPRESSOR
 /* ZSTD_initStats_ultra():
  * make a first compression pass, just to seed stats with more accurate starting values.
  * only works on first block, with no dictionary and no ldm.
  * this function cannot error out, its narrow contract must be respected.
  */
-static void
-ZSTD_initStats_ultra(ZSTD_matchState_t* ms,
-                     seqStore_t* seqStore,
-                     U32 rep[ZSTD_REP_NUM],
-               const void* src, size_t srcSize)
+static
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+void ZSTD_initStats_ultra(ZSTD_matchState_t* ms,
+                          seqStore_t* seqStore,
+                          U32 rep[ZSTD_REP_NUM],
+                    const void* src, size_t srcSize)
 {
     U32 tmpRep[ZSTD_REP_NUM];  /* updated rep codes will sink here */
     ZSTD_memcpy(tmpRep, rep, sizeof(tmpRep));
@@ -1423,7 +1524,7 @@ size_t ZSTD_compressBlock_btultra2(
      * Consequently, this can only work if no data has been previously loaded in tables,
      * aka, no dictionary, no prefix, no ldm preprocessing.
      * The compression ratio gain is generally small (~0.5% on first block),
-    ** the cost is 2x cpu time on first block. */
+     * the cost is 2x cpu time on first block. */
     assert(srcSize <= ZSTD_BLOCKSIZE_MAX);
     if ( (ms->opt.litLengthSum==0)   /* first block */
       && (seqStore->sequences == seqStore->sequencesStart)  /* no ldm */
@@ -1436,7 +1537,9 @@ size_t ZSTD_compressBlock_btultra2(
 
     return ZSTD_compressBlock_opt2(ms, seqStore, rep, src, srcSize, ZSTD_noDict);
 }
+#endif
 
+#ifndef ZSTD_EXCLUDE_BTOPT_BLOCK_COMPRESSOR
 size_t ZSTD_compressBlock_btopt_dictMatchState(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         const void* src, size_t srcSize)
@@ -1444,19 +1547,21 @@ size_t ZSTD_compressBlock_btopt_dictMatchState(
     return ZSTD_compressBlock_opt0(ms, seqStore, rep, src, srcSize, ZSTD_dictMatchState);
 }
 
-size_t ZSTD_compressBlock_btultra_dictMatchState(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        const void* src, size_t srcSize)
-{
-    return ZSTD_compressBlock_opt2(ms, seqStore, rep, src, srcSize, ZSTD_dictMatchState);
-}
-
 size_t ZSTD_compressBlock_btopt_extDict(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         const void* src, size_t srcSize)
 {
     return ZSTD_compressBlock_opt0(ms, seqStore, rep, src, srcSize, ZSTD_extDict);
 }
+#endif
+
+#ifndef ZSTD_EXCLUDE_BTULTRA_BLOCK_COMPRESSOR
+size_t ZSTD_compressBlock_btultra_dictMatchState(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        const void* src, size_t srcSize)
+{
+    return ZSTD_compressBlock_opt2(ms, seqStore, rep, src, srcSize, ZSTD_dictMatchState);
+}
 
 size_t ZSTD_compressBlock_btultra_extDict(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
@@ -1464,6 +1569,7 @@ size_t ZSTD_compressBlock_btultra_extDict(
 {
     return ZSTD_compressBlock_opt2(ms, seqStore, rep, src, srcSize, ZSTD_extDict);
 }
+#endif
 
 /* note : no btultra2 variant for extDict nor dictMatchState,
  * because btultra2 is not meant to work with dictionaries

lib/compress/zstd_opt.h

@@ -17,28 +17,38 @@ extern "C" {
 
 #include "zstd_compress_internal.h"
 
+#if !defined(ZSTD_EXCLUDE_BTLAZY2_BLOCK_COMPRESSOR) \
+ || !defined(ZSTD_EXCLUDE_BTOPT_BLOCK_COMPRESSOR) \
+ || !defined(ZSTD_EXCLUDE_BTULTRA_BLOCK_COMPRESSOR)
 /* used in ZSTD_loadDictionaryContent() */
 void ZSTD_updateTree(ZSTD_matchState_t* ms, const BYTE* ip, const BYTE* iend);
+#endif
 
+#ifndef ZSTD_EXCLUDE_BTOPT_BLOCK_COMPRESSOR
 size_t ZSTD_compressBlock_btopt(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
-size_t ZSTD_compressBlock_btultra(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize);
-size_t ZSTD_compressBlock_btultra2(
-        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
-        void const* src, size_t srcSize);
-
-
 size_t ZSTD_compressBlock_btopt_dictMatchState(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
-size_t ZSTD_compressBlock_btultra_dictMatchState(
+size_t ZSTD_compressBlock_btopt_extDict(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
 
-size_t ZSTD_compressBlock_btopt_extDict(
+#define ZSTD_COMPRESSBLOCK_BTOPT ZSTD_compressBlock_btopt
+#define ZSTD_COMPRESSBLOCK_BTOPT_DICTMATCHSTATE ZSTD_compressBlock_btopt_dictMatchState
+#define ZSTD_COMPRESSBLOCK_BTOPT_EXTDICT ZSTD_compressBlock_btopt_extDict
+#else
+#define ZSTD_COMPRESSBLOCK_BTOPT NULL
+#define ZSTD_COMPRESSBLOCK_BTOPT_DICTMATCHSTATE NULL
+#define ZSTD_COMPRESSBLOCK_BTOPT_EXTDICT NULL
+#endif
+
+#ifndef ZSTD_EXCLUDE_BTULTRA_BLOCK_COMPRESSOR
+size_t ZSTD_compressBlock_btultra(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+size_t ZSTD_compressBlock_btultra_dictMatchState(
         ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
         void const* src, size_t srcSize);
 size_t ZSTD_compressBlock_btultra_extDict(
@@ -48,6 +58,20 @@ size_t ZSTD_compressBlock_btultra_extDict(
         /* note : no btultra2 variant for extDict nor dictMatchState,
          * because btultra2 is not meant to work with dictionaries
          * and is only specific for the first block (no prefix) */
+size_t ZSTD_compressBlock_btultra2(
+        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
+        void const* src, size_t srcSize);
+
+#define ZSTD_COMPRESSBLOCK_BTULTRA ZSTD_compressBlock_btultra
+#define ZSTD_COMPRESSBLOCK_BTULTRA_DICTMATCHSTATE ZSTD_compressBlock_btultra_dictMatchState
+#define ZSTD_COMPRESSBLOCK_BTULTRA_EXTDICT ZSTD_compressBlock_btultra_extDict
+#define ZSTD_COMPRESSBLOCK_BTULTRA2 ZSTD_compressBlock_btultra2
+#else
+#define ZSTD_COMPRESSBLOCK_BTULTRA NULL
+#define ZSTD_COMPRESSBLOCK_BTULTRA_DICTMATCHSTATE NULL
+#define ZSTD_COMPRESSBLOCK_BTULTRA_EXTDICT NULL
+#define ZSTD_COMPRESSBLOCK_BTULTRA2 NULL
+#endif
 
 #if defined (__cplusplus)
 }

lib/compress/zstdmt_compress.c

@@ -15,16 +15,13 @@
 #endif
 
 
-/* ======   Constants   ====== */
-#define ZSTDMT_OVERLAPLOG_DEFAULT 0
-
-
 /* ======   Dependencies   ====== */
+#include "../common/allocations.h" /* ZSTD_customMalloc, ZSTD_customCalloc, ZSTD_customFree */
 #include "../common/zstd_deps.h"   /* ZSTD_memcpy, ZSTD_memset, INT_MAX, UINT_MAX */
 #include "../common/mem.h"         /* MEM_STATIC */
 #include "../common/pool.h"        /* threadpool */
 #include "../common/threading.h"   /* mutex */
-#include "zstd_compress_internal.h"  /* MIN, ERROR, ZSTD_*, ZSTD_highbit32 */
+#include "zstd_compress_internal.h" /* MIN, ERROR, ZSTD_*, ZSTD_highbit32 */
 #include "zstd_ldm.h"
 #include "zstdmt_compress.h"
 
@@ -43,12 +40,13 @@
 #  include <unistd.h>
 #  include <sys/times.h>
 
-#  define DEBUG_PRINTHEX(l,p,n) {            \
-    unsigned debug_u;                        \
-    for (debug_u=0; debug_u<(n); debug_u++)  \
-        RAWLOG(l, "%02X ", ((const unsigned char*)(p))[debug_u]); \
-    RAWLOG(l, " \n");                        \
-}
+#  define DEBUG_PRINTHEX(l,p,n)                                       \
+    do {                                                              \
+        unsigned debug_u;                                             \
+        for (debug_u=0; debug_u<(n); debug_u++)                       \
+            RAWLOG(l, "%02X ", ((const unsigned char*)(p))[debug_u]); \
+        RAWLOG(l, " \n");                                             \
+    } while (0)
 
 static unsigned long long GetCurrentClockTimeMicroseconds(void)
 {
@@ -60,25 +58,28 @@ static unsigned long long GetCurrentClockTimeMicroseconds(void)
 }  }
 
 #define MUTEX_WAIT_TIME_DLEVEL 6
-#define ZSTD_PTHREAD_MUTEX_LOCK(mutex) {          \
-    if (DEBUGLEVEL >= MUTEX_WAIT_TIME_DLEVEL) {   \
-        unsigned long long const beforeTime = GetCurrentClockTimeMicroseconds(); \
-        ZSTD_pthread_mutex_lock(mutex);           \
-        {   unsigned long long const afterTime = GetCurrentClockTimeMicroseconds(); \
-            unsigned long long const elapsedTime = (afterTime-beforeTime); \
-            if (elapsedTime > 1000) {  /* or whatever threshold you like; I'm using 1 millisecond here */ \
-                DEBUGLOG(MUTEX_WAIT_TIME_DLEVEL, "Thread took %llu microseconds to acquire mutex %s \n", \
-                   elapsedTime, #mutex);          \
-        }   }                                     \
-    } else {                                      \
-        ZSTD_pthread_mutex_lock(mutex);           \
-    }                                             \
-}
+#define ZSTD_PTHREAD_MUTEX_LOCK(mutex)                                                  \
+    do {                                                                                \
+        if (DEBUGLEVEL >= MUTEX_WAIT_TIME_DLEVEL) {                                     \
+            unsigned long long const beforeTime = GetCurrentClockTimeMicroseconds();    \
+            ZSTD_pthread_mutex_lock(mutex);                                             \
+            {   unsigned long long const afterTime = GetCurrentClockTimeMicroseconds(); \
+                unsigned long long const elapsedTime = (afterTime-beforeTime);          \
+                if (elapsedTime > 1000) {                                               \
+                    /* or whatever threshold you like; I'm using 1 millisecond here */  \
+                    DEBUGLOG(MUTEX_WAIT_TIME_DLEVEL,                                    \
+                        "Thread took %llu microseconds to acquire mutex %s \n",         \
+                        elapsedTime, #mutex);                                           \
+            }   }                                                                       \
+        } else {                                                                        \
+            ZSTD_pthread_mutex_lock(mutex);                                             \
+        }                                                                               \
+    } while (0)
 
 #else
 
 #  define ZSTD_PTHREAD_MUTEX_LOCK(m) ZSTD_pthread_mutex_lock(m)
-#  define DEBUG_PRINTHEX(l,p,n) {}
+#  define DEBUG_PRINTHEX(l,p,n) do { } while (0)
 
 #endif
 
@@ -99,18 +100,39 @@ typedef struct ZSTDMT_bufferPool_s {
     unsigned totalBuffers;
     unsigned nbBuffers;
     ZSTD_customMem cMem;
-    buffer_t bTable[1];   /* variable size */
+    buffer_t* buffers;
 } ZSTDMT_bufferPool;
 
+static void ZSTDMT_freeBufferPool(ZSTDMT_bufferPool* bufPool)
+{
+    DEBUGLOG(3, "ZSTDMT_freeBufferPool (address:%08X)", (U32)(size_t)bufPool);
+    if (!bufPool) return;   /* compatibility with free on NULL */
+    if (bufPool->buffers) {
+        unsigned u;
+        for (u=0; u<bufPool->totalBuffers; u++) {
+            DEBUGLOG(4, "free buffer %2u (address:%08X)", u, (U32)(size_t)bufPool->buffers[u].start);
+            ZSTD_customFree(bufPool->buffers[u].start, bufPool->cMem);
+        }
+        ZSTD_customFree(bufPool->buffers, bufPool->cMem);
+    }
+    ZSTD_pthread_mutex_destroy(&bufPool->poolMutex);
+    ZSTD_customFree(bufPool, bufPool->cMem);
+}
+
 static ZSTDMT_bufferPool* ZSTDMT_createBufferPool(unsigned maxNbBuffers, ZSTD_customMem cMem)
 {
-    ZSTDMT_bufferPool* const bufPool = (ZSTDMT_bufferPool*)ZSTD_customCalloc(
-        sizeof(ZSTDMT_bufferPool) + (maxNbBuffers-1) * sizeof(buffer_t), cMem);
+    ZSTDMT_bufferPool* const bufPool =
+        (ZSTDMT_bufferPool*)ZSTD_customCalloc(sizeof(ZSTDMT_bufferPool), cMem);
     if (bufPool==NULL) return NULL;
     if (ZSTD_pthread_mutex_init(&bufPool->poolMutex, NULL)) {
         ZSTD_customFree(bufPool, cMem);
         return NULL;
     }
+    bufPool->buffers = (buffer_t*)ZSTD_customCalloc(maxNbBuffers * sizeof(buffer_t), cMem);
+    if (bufPool->buffers==NULL) {
+        ZSTDMT_freeBufferPool(bufPool);
+        return NULL;
+    }
     bufPool->bufferSize = 64 KB;
     bufPool->totalBuffers = maxNbBuffers;
     bufPool->nbBuffers = 0;
@@ -118,32 +140,19 @@ static ZSTDMT_bufferPool* ZSTDMT_createBufferPool(unsigned maxNbBuffers, ZSTD_cu
     return bufPool;
 }
 
-static void ZSTDMT_freeBufferPool(ZSTDMT_bufferPool* bufPool)
-{
-    unsigned u;
-    DEBUGLOG(3, "ZSTDMT_freeBufferPool (address:%08X)", (U32)(size_t)bufPool);
-    if (!bufPool) return;   /* compatibility with free on NULL */
-    for (u=0; u<bufPool->totalBuffers; u++) {
-        DEBUGLOG(4, "free buffer %2u (address:%08X)", u, (U32)(size_t)bufPool->bTable[u].start);
-        ZSTD_customFree(bufPool->bTable[u].start, bufPool->cMem);
-    }
-    ZSTD_pthread_mutex_destroy(&bufPool->poolMutex);
-    ZSTD_customFree(bufPool, bufPool->cMem);
-}
-
 /* only works at initialization, not during compression */
 static size_t ZSTDMT_sizeof_bufferPool(ZSTDMT_bufferPool* bufPool)
 {
-    size_t const poolSize = sizeof(*bufPool)
-                          + (bufPool->totalBuffers - 1) * sizeof(buffer_t);
+    size_t const poolSize = sizeof(*bufPool);
+    size_t const arraySize = bufPool->totalBuffers * sizeof(buffer_t);
     unsigned u;
     size_t totalBufferSize = 0;
     ZSTD_pthread_mutex_lock(&bufPool->poolMutex);
     for (u=0; u<bufPool->totalBuffers; u++)
-        totalBufferSize += bufPool->bTable[u].capacity;
+        totalBufferSize += bufPool->buffers[u].capacity;
     ZSTD_pthread_mutex_unlock(&bufPool->poolMutex);
 
-    return poolSize + totalBufferSize;
+    return poolSize + arraySize + totalBufferSize;
 }
 
 /* ZSTDMT_setBufferSize() :
@@ -186,9 +195,9 @@ static buffer_t ZSTDMT_getBuffer(ZSTDMT_bufferPool* bufPool)
     DEBUGLOG(5, "ZSTDMT_getBuffer: bSize = %u", (U32)bufPool->bufferSize);
     ZSTD_pthread_mutex_lock(&bufPool->poolMutex);
     if (bufPool->nbBuffers) {   /* try to use an existing buffer */
-        buffer_t const buf = bufPool->bTable[--(bufPool->nbBuffers)];
+        buffer_t const buf = bufPool->buffers[--(bufPool->nbBuffers)];
         size_t const availBufferSize = buf.capacity;
-        bufPool->bTable[bufPool->nbBuffers] = g_nullBuffer;
+        bufPool->buffers[bufPool->nbBuffers] = g_nullBuffer;
         if ((availBufferSize >= bSize) & ((availBufferSize>>3) <= bSize)) {
             /* large enough, but not too much */
             DEBUGLOG(5, "ZSTDMT_getBuffer: provide buffer %u of size %u",
@@ -249,14 +258,14 @@ static void ZSTDMT_releaseBuffer(ZSTDMT_bufferPool* bufPool, buffer_t buf)
     if (buf.start == NULL) return;   /* compatible with release on NULL */
     ZSTD_pthread_mutex_lock(&bufPool->poolMutex);
     if (bufPool->nbBuffers < bufPool->totalBuffers) {
-        bufPool->bTable[bufPool->nbBuffers++] = buf;  /* stored for later use */
+        bufPool->buffers[bufPool->nbBuffers++] = buf;  /* stored for later use */
         DEBUGLOG(5, "ZSTDMT_releaseBuffer: stored buffer of size %u in slot %u",
                     (U32)buf.capacity, (U32)(bufPool->nbBuffers-1));
         ZSTD_pthread_mutex_unlock(&bufPool->poolMutex);
         return;
     }
     ZSTD_pthread_mutex_unlock(&bufPool->poolMutex);
-    /* Reached bufferPool capacity (should not happen) */
+    /* Reached bufferPool capacity (note: should not happen) */
     DEBUGLOG(5, "ZSTDMT_releaseBuffer: pool capacity reached => freeing ");
     ZSTD_customFree(buf.start, bufPool->cMem);
 }
@@ -349,16 +358,20 @@ typedef struct {
     int totalCCtx;
     int availCCtx;
     ZSTD_customMem cMem;
-    ZSTD_CCtx* cctx[1];   /* variable size */
+    ZSTD_CCtx** cctxs;
 } ZSTDMT_CCtxPool;
 
-/* note : all CCtx borrowed from the pool should be released back to the pool _before_ freeing the pool */
+/* note : all CCtx borrowed from the pool must be reverted back to the pool _before_ freeing the pool */
 static void ZSTDMT_freeCCtxPool(ZSTDMT_CCtxPool* pool)
 {
-    int cid;
-    for (cid=0; cid<pool->totalCCtx; cid++)
-        ZSTD_freeCCtx(pool->cctx[cid]);  /* note : compatible with free on NULL */
+    if (!pool) return;
     ZSTD_pthread_mutex_destroy(&pool->poolMutex);
+    if (pool->cctxs) {
+        int cid;
+        for (cid=0; cid<pool->totalCCtx; cid++)
+            ZSTD_freeCCtx(pool->cctxs[cid]);  /* free compatible with NULL */
+        ZSTD_customFree(pool->cctxs, pool->cMem);
+    }
     ZSTD_customFree(pool, pool->cMem);
 }
 
@@ -367,19 +380,24 @@ static void ZSTDMT_freeCCtxPool(ZSTDMT_CCtxPool* pool)
 static ZSTDMT_CCtxPool* ZSTDMT_createCCtxPool(int nbWorkers,
                                               ZSTD_customMem cMem)
 {
-    ZSTDMT_CCtxPool* const cctxPool = (ZSTDMT_CCtxPool*) ZSTD_customCalloc(
-        sizeof(ZSTDMT_CCtxPool) + (nbWorkers-1)*sizeof(ZSTD_CCtx*), cMem);
+    ZSTDMT_CCtxPool* const cctxPool =
+        (ZSTDMT_CCtxPool*) ZSTD_customCalloc(sizeof(ZSTDMT_CCtxPool), cMem);
     assert(nbWorkers > 0);
     if (!cctxPool) return NULL;
     if (ZSTD_pthread_mutex_init(&cctxPool->poolMutex, NULL)) {
         ZSTD_customFree(cctxPool, cMem);
         return NULL;
     }
-    cctxPool->cMem = cMem;
     cctxPool->totalCCtx = nbWorkers;
+    cctxPool->cctxs = (ZSTD_CCtx**)ZSTD_customCalloc(nbWorkers * sizeof(ZSTD_CCtx*), cMem);
+    if (!cctxPool->cctxs) {
+        ZSTDMT_freeCCtxPool(cctxPool);
+        return NULL;
+    }
+    cctxPool->cMem = cMem;
+    cctxPool->cctxs[0] = ZSTD_createCCtx_advanced(cMem);
+    if (!cctxPool->cctxs[0]) { ZSTDMT_freeCCtxPool(cctxPool); return NULL; }
     cctxPool->availCCtx = 1;   /* at least one cctx for single-thread mode */
-    cctxPool->cctx[0] = ZSTD_createCCtx_advanced(cMem);
-    if (!cctxPool->cctx[0]) { ZSTDMT_freeCCtxPool(cctxPool); return NULL; }
     DEBUGLOG(3, "cctxPool created, with %u workers", nbWorkers);
     return cctxPool;
 }
@@ -401,16 +419,16 @@ static size_t ZSTDMT_sizeof_CCtxPool(ZSTDMT_CCtxPool* cctxPool)
 {
     ZSTD_pthread_mutex_lock(&cctxPool->poolMutex);
     {   unsigned const nbWorkers = cctxPool->totalCCtx;
-        size_t const poolSize = sizeof(*cctxPool)
-                                + (nbWorkers-1) * sizeof(ZSTD_CCtx*);
-        unsigned u;
+        size_t const poolSize = sizeof(*cctxPool);
+        size_t const arraySize = cctxPool->totalCCtx * sizeof(ZSTD_CCtx*);
         size_t totalCCtxSize = 0;
+        unsigned u;
         for (u=0; u<nbWorkers; u++) {
-            totalCCtxSize += ZSTD_sizeof_CCtx(cctxPool->cctx[u]);
+            totalCCtxSize += ZSTD_sizeof_CCtx(cctxPool->cctxs[u]);
         }
         ZSTD_pthread_mutex_unlock(&cctxPool->poolMutex);
         assert(nbWorkers > 0);
-        return poolSize + totalCCtxSize;
+        return poolSize + arraySize + totalCCtxSize;
     }
 }
 
@@ -420,7 +438,7 @@ static ZSTD_CCtx* ZSTDMT_getCCtx(ZSTDMT_CCtxPool* cctxPool)
     ZSTD_pthread_mutex_lock(&cctxPool->poolMutex);
     if (cctxPool->availCCtx) {
         cctxPool->availCCtx--;
-        {   ZSTD_CCtx* const cctx = cctxPool->cctx[cctxPool->availCCtx];
+        {   ZSTD_CCtx* const cctx = cctxPool->cctxs[cctxPool->availCCtx];
             ZSTD_pthread_mutex_unlock(&cctxPool->poolMutex);
             return cctx;
     }   }
@@ -434,7 +452,7 @@ static void ZSTDMT_releaseCCtx(ZSTDMT_CCtxPool* pool, ZSTD_CCtx* cctx)
     if (cctx==NULL) return;   /* compatibility with release on NULL */
     ZSTD_pthread_mutex_lock(&pool->poolMutex);
     if (pool->availCCtx < pool->totalCCtx)
-        pool->cctx[pool->availCCtx++] = cctx;
+        pool->cctxs[pool->availCCtx++] = cctx;
     else {
         /* pool overflow : should not happen, since totalCCtx==nbWorkers */
         DEBUGLOG(4, "CCtx pool overflow : free cctx");
@@ -600,11 +618,8 @@ static void ZSTDMT_serialState_update(serialState_t* serialState,
     ZSTD_pthread_mutex_unlock(&serialState->mutex);
 
     if (seqStore.size > 0) {
-        size_t const err = ZSTD_referenceExternalSequences(
-            jobCCtx, seqStore.seq, seqStore.size);
+        ZSTD_referenceExternalSequences(jobCCtx, seqStore.seq, seqStore.size);
         assert(serialState->params.ldmParams.enableLdm == ZSTD_ps_enable);
-        assert(!ZSTD_isError(err));
-        (void)err;
     }
 }
 
@@ -656,12 +671,13 @@ typedef struct {
     unsigned frameChecksumNeeded;        /* used only by mtctx */
 } ZSTDMT_jobDescription;
 
-#define JOB_ERROR(e) {                          \
-    ZSTD_PTHREAD_MUTEX_LOCK(&job->job_mutex);   \
-    job->cSize = e;                             \
-    ZSTD_pthread_mutex_unlock(&job->job_mutex); \
-    goto _endJob;                               \
-}
+#define JOB_ERROR(e)                                \
+    do {                                            \
+        ZSTD_PTHREAD_MUTEX_LOCK(&job->job_mutex);   \
+        job->cSize = e;                             \
+        ZSTD_pthread_mutex_unlock(&job->job_mutex); \
+        goto _endJob;                               \
+    } while (0)
 
 /* ZSTDMT_compressionJob() is a POOL_function type */
 static void ZSTDMT_compressionJob(void* jobDescription)
@@ -719,7 +735,7 @@ static void ZSTDMT_compressionJob(void* jobDescription)
     ZSTDMT_serialState_update(job->serial, cctx, rawSeqStore, job->src, job->jobID);
 
     if (!job->firstJob) {  /* flush and overwrite frame header when it's not first job */
-        size_t const hSize = ZSTD_compressContinue(cctx, dstBuff.start, dstBuff.capacity, job->src.start, 0);
+        size_t const hSize = ZSTD_compressContinue_public(cctx, dstBuff.start, dstBuff.capacity, job->src.start, 0);
         if (ZSTD_isError(hSize)) JOB_ERROR(hSize);
         DEBUGLOG(5, "ZSTDMT_compressionJob: flush and overwrite %u bytes of frame header (not first job)", (U32)hSize);
         ZSTD_invalidateRepCodes(cctx);
@@ -737,7 +753,7 @@ static void ZSTDMT_compressionJob(void* jobDescription)
         DEBUGLOG(5, "ZSTDMT_compressionJob: compress %u bytes in %i blocks", (U32)job->src.size, nbChunks);
         assert(job->cSize == 0);
         for (chunkNb = 1; chunkNb < nbChunks; chunkNb++) {
-            size_t const cSize = ZSTD_compressContinue(cctx, op, oend-op, ip, chunkSize);
+            size_t const cSize = ZSTD_compressContinue_public(cctx, op, oend-op, ip, chunkSize);
             if (ZSTD_isError(cSize)) JOB_ERROR(cSize);
             ip += chunkSize;
             op += cSize; assert(op < oend);
@@ -757,8 +773,8 @@ static void ZSTDMT_compressionJob(void* jobDescription)
             size_t const lastBlockSize1 = job->src.size & (chunkSize-1);
             size_t const lastBlockSize = ((lastBlockSize1==0) & (job->src.size>=chunkSize)) ? chunkSize : lastBlockSize1;
             size_t const cSize = (job->lastJob) ?
-                 ZSTD_compressEnd     (cctx, op, oend-op, ip, lastBlockSize) :
-                 ZSTD_compressContinue(cctx, op, oend-op, ip, lastBlockSize);
+                 ZSTD_compressEnd_public(cctx, op, oend-op, ip, lastBlockSize) :
+                 ZSTD_compressContinue_public(cctx, op, oend-op, ip, lastBlockSize);
             if (ZSTD_isError(cSize)) JOB_ERROR(cSize);
             lastCBlockSize = cSize;
     }   }
@@ -1090,7 +1106,7 @@ ZSTD_frameProgression ZSTDMT_getFrameProgression(ZSTDMT_CCtx* mtctx)
     {   unsigned jobNb;
         unsigned lastJobNb = mtctx->nextJobID + mtctx->jobReady; assert(mtctx->jobReady <= 1);
         DEBUGLOG(6, "ZSTDMT_getFrameProgression: jobs: from %u to <%u (jobReady:%u)",
-                    mtctx->doneJobID, lastJobNb, mtctx->jobReady)
+                    mtctx->doneJobID, lastJobNb, mtctx->jobReady);
         for (jobNb = mtctx->doneJobID ; jobNb < lastJobNb ; jobNb++) {
             unsigned const wJobID = jobNb & mtctx->jobIDMask;
             ZSTDMT_jobDescription* jobPtr = &mtctx->jobs[wJobID];

lib/decompress/huf_decompress.c

@@ -34,6 +34,12 @@
 *  Macros
 ****************************************************************/
 
+#ifdef HUF_DISABLE_FAST_DECODE
+# define HUF_ENABLE_FAST_DECODE 0
+#else
+# define HUF_ENABLE_FAST_DECODE 1
+#endif
+
 /* These two optional macros force the use one way or another of the two
  * Huffman decompression implementations. You can't force in both directions
  * at the same time.
@@ -158,17 +164,18 @@ static size_t HUF_initFastDStream(BYTE const* ip) {
  * op [in/out] - The output pointers, must be updated to reflect what is written.
  * bits [in/out] - The bitstream containers, must be updated to reflect the current state.
  * dt [in] - The decoding table.
- * ilimit [in] - The input limit, stop when any input pointer is below ilimit.
+ * ilowest [in] - The beginning of the valid range of the input. Decoders may read
+ *                down to this pointer. It may be below iend[0].
  * oend [in] - The end of the output stream. op[3] must not cross oend.
  * iend [in] - The end of each input stream. ip[i] may cross iend[i],
- *             as long as it is above ilimit, but that indicates corruption.
+ *             as long as it is above ilowest, but that indicates corruption.
  */
 typedef struct {
     BYTE const* ip[4];
     BYTE* op[4];
     U64 bits[4];
     void const* dt;
-    BYTE const* ilimit;
+    BYTE const* ilowest;
     BYTE* oend;
     BYTE const* iend[4];
 } HUF_DecompressFastArgs;
@@ -186,9 +193,9 @@ static size_t HUF_DecompressFastArgs_init(HUF_DecompressFastArgs* args, void* ds
     void const* dt = DTable + 1;
     U32 const dtLog = HUF_getDTableDesc(DTable).tableLog;
 
-    const BYTE* const ilimit = (const BYTE*)src + 6 + 8;
+    const BYTE* const istart = (const BYTE*)src;
 
-    BYTE* const oend = (BYTE*)dst + dstSize;
+    BYTE* const oend = ZSTD_maybeNullPtrAdd((BYTE*)dst, dstSize);
 
     /* The fast decoding loop assumes 64-bit little-endian.
      * This condition is false on x32.
@@ -196,6 +203,11 @@ static size_t HUF_DecompressFastArgs_init(HUF_DecompressFastArgs* args, void* ds
     if (!MEM_isLittleEndian() || MEM_32bits())
         return 0;
 
+    /* Avoid nullptr addition */
+    if (dstSize == 0)
+        return 0;
+    assert(dst != NULL);
+
     /* strict minimum : jump table + 1 byte per stream */
     if (srcSize < 10)
         return ERROR(corruption_detected);
@@ -209,7 +221,6 @@ static size_t HUF_DecompressFastArgs_init(HUF_DecompressFastArgs* args, void* ds
 
     /* Read the jump table. */
     {
-        const BYTE* const istart = (const BYTE*)src;
         size_t const length1 = MEM_readLE16(istart);
         size_t const length2 = MEM_readLE16(istart+2);
         size_t const length3 = MEM_readLE16(istart+4);
@@ -221,10 +232,8 @@ static size_t HUF_DecompressFastArgs_init(HUF_DecompressFastArgs* args, void* ds
 
         /* HUF_initFastDStream() requires this, and this small of an input
          * won't benefit from the ASM loop anyways.
-         * length1 must be >= 16 so that ip[0] >= ilimit before the loop
-         * starts.
          */
-        if (length1 < 16 || length2 < 8 || length3 < 8 || length4 < 8)
+        if (length1 < 8 || length2 < 8 || length3 < 8 || length4 < 8)
             return 0;
         if (length4 > srcSize) return ERROR(corruption_detected);   /* overflow */
     }
@@ -256,11 +265,12 @@ static size_t HUF_DecompressFastArgs_init(HUF_DecompressFastArgs* args, void* ds
     args->bits[2] = HUF_initFastDStream(args->ip[2]);
     args->bits[3] = HUF_initFastDStream(args->ip[3]);
 
-    /* If ip[] >= ilimit, it is guaranteed to be safe to
-        * reload bits[]. It may be beyond its section, but is
-        * guaranteed to be valid (>= istart).
-        */
-    args->ilimit = ilimit;
+    /* The decoders must be sure to never read beyond ilowest.
+     * This is lower than iend[0], but allowing decoders to read
+     * down to ilowest can allow an extra iteration or two in the
+     * fast loop.
+     */
+    args->ilowest = istart;
 
     args->oend = oend;
     args->dt = dt;
@@ -285,13 +295,31 @@ static size_t HUF_initRemainingDStream(BIT_DStream_t* bit, HUF_DecompressFastArg
     assert(sizeof(size_t) == 8);
     bit->bitContainer = MEM_readLEST(args->ip[stream]);
     bit->bitsConsumed = ZSTD_countTrailingZeros64(args->bits[stream]);
-    bit->start = (const char*)args->iend[0];
+    bit->start = (const char*)args->ilowest;
     bit->limitPtr = bit->start + sizeof(size_t);
     bit->ptr = (const char*)args->ip[stream];
 
     return 0;
 }
 
+/* Calls X(N) for each stream 0, 1, 2, 3. */
+#define HUF_4X_FOR_EACH_STREAM(X) \
+    do {                          \
+        X(0);                     \
+        X(1);                     \
+        X(2);                     \
+        X(3);                     \
+    } while (0)
+
+/* Calls X(N, var) for each stream 0, 1, 2, 3. */
+#define HUF_4X_FOR_EACH_STREAM_WITH_VAR(X, var) \
+    do {                                        \
+        X(0, (var));                            \
+        X(1, (var));                            \
+        X(2, (var));                            \
+        X(3, (var));                            \
+    } while (0)
+
 
 #ifndef HUF_FORCE_DECOMPRESS_X2
 
@@ -500,15 +528,19 @@ HUF_decodeSymbolX1(BIT_DStream_t* Dstream, const HUF_DEltX1* dt, const U32 dtLog
 }
 
 #define HUF_DECODE_SYMBOLX1_0(ptr, DStreamPtr) \
-    *ptr++ = HUF_decodeSymbolX1(DStreamPtr, dt, dtLog)
+    do { *ptr++ = HUF_decodeSymbolX1(DStreamPtr, dt, dtLog); } while (0)
 
-#define HUF_DECODE_SYMBOLX1_1(ptr, DStreamPtr)  \
-    if (MEM_64bits() || (HUF_TABLELOG_MAX<=12)) \
-        HUF_DECODE_SYMBOLX1_0(ptr, DStreamPtr)
+#define HUF_DECODE_SYMBOLX1_1(ptr, DStreamPtr)      \
+    do {                                            \
+        if (MEM_64bits() || (HUF_TABLELOG_MAX<=12)) \
+            HUF_DECODE_SYMBOLX1_0(ptr, DStreamPtr); \
+    } while (0)
 
-#define HUF_DECODE_SYMBOLX1_2(ptr, DStreamPtr) \
-    if (MEM_64bits()) \
-        HUF_DECODE_SYMBOLX1_0(ptr, DStreamPtr)
+#define HUF_DECODE_SYMBOLX1_2(ptr, DStreamPtr)      \
+    do {                                            \
+        if (MEM_64bits())                           \
+            HUF_DECODE_SYMBOLX1_0(ptr, DStreamPtr); \
+    } while (0)
 
 HINT_INLINE size_t
 HUF_decodeStreamX1(BYTE* p, BIT_DStream_t* const bitDPtr, BYTE* const pEnd, const HUF_DEltX1* const dt, const U32 dtLog)
@@ -546,7 +578,7 @@ HUF_decompress1X1_usingDTable_internal_body(
     const HUF_DTable* DTable)
 {
     BYTE* op = (BYTE*)dst;
-    BYTE* const oend = op + dstSize;
+    BYTE* const oend = ZSTD_maybeNullPtrAdd(op, dstSize);
     const void* dtPtr = DTable + 1;
     const HUF_DEltX1* const dt = (const HUF_DEltX1*)dtPtr;
     BIT_DStream_t bitD;
@@ -574,6 +606,7 @@ HUF_decompress4X1_usingDTable_internal_body(
 {
     /* Check */
     if (cSrcSize < 10) return ERROR(corruption_detected);  /* strict minimum : jump table + 1 byte per stream */
+    if (dstSize < 6) return ERROR(corruption_detected);         /* stream 4-split doesn't work */
 
     {   const BYTE* const istart = (const BYTE*) cSrc;
         BYTE* const ostart = (BYTE*) dst;
@@ -609,7 +642,7 @@ HUF_decompress4X1_usingDTable_internal_body(
 
         if (length4 > cSrcSize) return ERROR(corruption_detected);   /* overflow */
         if (opStart4 > oend) return ERROR(corruption_detected);      /* overflow */
-        if (dstSize < 6) return ERROR(corruption_detected);         /* stream 4-split doesn't work */
+        assert(dstSize >= 6); /* validated above */
         CHECK_F( BIT_initDStream(&bitD1, istart1, length1) );
         CHECK_F( BIT_initDStream(&bitD2, istart2, length2) );
         CHECK_F( BIT_initDStream(&bitD3, istart3, length3) );
@@ -692,11 +725,11 @@ void HUF_decompress4X1_usingDTable_internal_fast_c_loop(HUF_DecompressFastArgs*
     BYTE* op[4];
     U16 const* const dtable = (U16 const*)args->dt;
     BYTE* const oend = args->oend;
-    BYTE const* const ilimit = args->ilimit;
+    BYTE const* const ilowest = args->ilowest;
 
     /* Copy the arguments to local variables */
     ZSTD_memcpy(&bits, &args->bits, sizeof(bits));
-    ZSTD_memcpy(&ip, &args->ip, sizeof(ip));
+    ZSTD_memcpy((void*)(&ip), &args->ip, sizeof(ip));
     ZSTD_memcpy(&op, &args->op, sizeof(op));
 
     assert(MEM_isLittleEndian());
@@ -705,13 +738,12 @@ void HUF_decompress4X1_usingDTable_internal_fast_c_loop(HUF_DecompressFastArgs*
     for (;;) {
         BYTE* olimit;
         int stream;
-        int symbol;
 
         /* Assert loop preconditions */
 #ifndef NDEBUG
         for (stream = 0; stream < 4; ++stream) {
             assert(op[stream] <= (stream == 3 ? oend : op[stream + 1]));
-            assert(ip[stream] >= ilimit);
+            assert(ip[stream] >= ilowest);
         }
 #endif
         /* Compute olimit */
@@ -721,7 +753,7 @@ void HUF_decompress4X1_usingDTable_internal_fast_c_loop(HUF_DecompressFastArgs*
             /* Each iteration consumes up to 11 bits * 5 = 55 bits < 7 bytes
              * per stream.
              */
-            size_t const iiters = (size_t)(ip[0] - ilimit) / 7;
+            size_t const iiters = (size_t)(ip[0] - ilowest) / 7;
             /* We can safely run iters iterations before running bounds checks */
             size_t const iters = MIN(oiters, iiters);
             size_t const symbols = iters * 5;
@@ -732,8 +764,8 @@ void HUF_decompress4X1_usingDTable_internal_fast_c_loop(HUF_DecompressFastArgs*
              */
             olimit = op[3] + symbols;
 
-            /* Exit fast decoding loop once we get close to the end. */
-            if (op[3] + 20 > olimit)
+            /* Exit fast decoding loop once we reach the end. */
+            if (op[3] == olimit)
                 break;
 
             /* Exit the decoding loop if any input pointer has crossed the
@@ -752,34 +784,49 @@ void HUF_decompress4X1_usingDTable_internal_fast_c_loop(HUF_DecompressFastArgs*
         }
 #endif
 
+#define HUF_4X1_DECODE_SYMBOL(_stream, _symbol)                 \
+    do {                                                        \
+        int const index = (int)(bits[(_stream)] >> 53);         \
+        int const entry = (int)dtable[index];                   \
+        bits[(_stream)] <<= (entry & 0x3F);                     \
+        op[(_stream)][(_symbol)] = (BYTE)((entry >> 8) & 0xFF); \
+    } while (0)
+
+#define HUF_4X1_RELOAD_STREAM(_stream)                              \
+    do {                                                            \
+        int const ctz = ZSTD_countTrailingZeros64(bits[(_stream)]); \
+        int const nbBits = ctz & 7;                                 \
+        int const nbBytes = ctz >> 3;                               \
+        op[(_stream)] += 5;                                         \
+        ip[(_stream)] -= nbBytes;                                   \
+        bits[(_stream)] = MEM_read64(ip[(_stream)]) | 1;            \
+        bits[(_stream)] <<= nbBits;                                 \
+    } while (0)
+
+        /* Manually unroll the loop because compilers don't consistently
+         * unroll the inner loops, which destroys performance.
+         */
         do {
             /* Decode 5 symbols in each of the 4 streams */
-            for (symbol = 0; symbol < 5; ++symbol) {
-                for (stream = 0; stream < 4; ++stream) {
-                    int const index = (int)(bits[stream] >> 53);
-                    int const entry = (int)dtable[index];
-                    bits[stream] <<= (entry & 63);
-                    op[stream][symbol] = (BYTE)((entry >> 8) & 0xFF);
-                }
-            }
-            /* Reload the bitstreams */
-            for (stream = 0; stream < 4; ++stream) {
-                int const ctz = ZSTD_countTrailingZeros64(bits[stream]);
-                int const nbBits = ctz & 7;
-                int const nbBytes = ctz >> 3;
-                op[stream] += 5;
-                ip[stream] -= nbBytes;
-                bits[stream] = MEM_read64(ip[stream]) | 1;
-                bits[stream] <<= nbBits;
-            }
+            HUF_4X_FOR_EACH_STREAM_WITH_VAR(HUF_4X1_DECODE_SYMBOL, 0);
+            HUF_4X_FOR_EACH_STREAM_WITH_VAR(HUF_4X1_DECODE_SYMBOL, 1);
+            HUF_4X_FOR_EACH_STREAM_WITH_VAR(HUF_4X1_DECODE_SYMBOL, 2);
+            HUF_4X_FOR_EACH_STREAM_WITH_VAR(HUF_4X1_DECODE_SYMBOL, 3);
+            HUF_4X_FOR_EACH_STREAM_WITH_VAR(HUF_4X1_DECODE_SYMBOL, 4);
+
+            /* Reload each of the 4 the bitstreams */
+            HUF_4X_FOR_EACH_STREAM(HUF_4X1_RELOAD_STREAM);
         } while (op[3] < olimit);
+
+#undef HUF_4X1_DECODE_SYMBOL
+#undef HUF_4X1_RELOAD_STREAM
     }
 
 _out:
 
     /* Save the final values of each of the state variables back to args. */
     ZSTD_memcpy(&args->bits, &bits, sizeof(bits));
-    ZSTD_memcpy(&args->ip, &ip, sizeof(ip));
+    ZSTD_memcpy((void*)(&args->ip), &ip, sizeof(ip));
     ZSTD_memcpy(&args->op, &op, sizeof(op));
 }
 
@@ -797,8 +844,8 @@ HUF_decompress4X1_usingDTable_internal_fast(
     HUF_DecompressFastLoopFn loopFn)
 {
     void const* dt = DTable + 1;
-    const BYTE* const iend = (const BYTE*)cSrc + 6;
-    BYTE* const oend = (BYTE*)dst + dstSize;
+    BYTE const* const ilowest = (BYTE const*)cSrc;
+    BYTE* const oend = ZSTD_maybeNullPtrAdd((BYTE*)dst, dstSize);
     HUF_DecompressFastArgs args;
     {   size_t const ret = HUF_DecompressFastArgs_init(&args, dst, dstSize, cSrc, cSrcSize, DTable);
         FORWARD_IF_ERROR(ret, "Failed to init fast loop args");
@@ -806,18 +853,22 @@ HUF_decompress4X1_usingDTable_internal_fast(
             return 0;
     }
 
-    assert(args.ip[0] >= args.ilimit);
+    assert(args.ip[0] >= args.ilowest);
     loopFn(&args);
 
-    /* Our loop guarantees that ip[] >= ilimit and that we haven't
+    /* Our loop guarantees that ip[] >= ilowest and that we haven't
     * overwritten any op[].
     */
-    assert(args.ip[0] >= iend);
-    assert(args.ip[1] >= iend);
-    assert(args.ip[2] >= iend);
-    assert(args.ip[3] >= iend);
+    assert(args.ip[0] >= ilowest);
+    assert(args.ip[0] >= ilowest);
+    assert(args.ip[1] >= ilowest);
+    assert(args.ip[2] >= ilowest);
+    assert(args.ip[3] >= ilowest);
     assert(args.op[3] <= oend);
-    (void)iend;
+
+    assert(ilowest == args.ilowest);
+    assert(ilowest + 6 == args.iend[0]);
+    (void)ilowest;
 
     /* finish bit streams one by one. */
     {   size_t const segmentSize = (dstSize+3) / 4;
@@ -868,7 +919,7 @@ static size_t HUF_decompress4X1_usingDTable_internal(void* dst, size_t dstSize,
     }
 #endif
 
-    if (!(flags & HUF_flags_disableFast)) {
+    if (HUF_ENABLE_FAST_DECODE && !(flags & HUF_flags_disableFast)) {
         size_t const ret = HUF_decompress4X1_usingDTable_internal_fast(dst, dstSize, cSrc, cSrcSize, DTable, loopFn);
         if (ret != 0)
             return ret;
@@ -1239,15 +1290,19 @@ HUF_decodeLastSymbolX2(void* op, BIT_DStream_t* DStream, const HUF_DEltX2* dt, c
 }
 
 #define HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr) \
-    ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog)
+    do { ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog); } while (0)
 
-#define HUF_DECODE_SYMBOLX2_1(ptr, DStreamPtr) \
-    if (MEM_64bits() || (HUF_TABLELOG_MAX<=12)) \
-        ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog)
+#define HUF_DECODE_SYMBOLX2_1(ptr, DStreamPtr)                     \
+    do {                                                           \
+        if (MEM_64bits() || (HUF_TABLELOG_MAX<=12))                \
+            ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog); \
+    } while (0)
 
-#define HUF_DECODE_SYMBOLX2_2(ptr, DStreamPtr) \
-    if (MEM_64bits()) \
-        ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog)
+#define HUF_DECODE_SYMBOLX2_2(ptr, DStreamPtr)                     \
+    do {                                                           \
+        if (MEM_64bits())                                          \
+            ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog); \
+    } while (0)
 
 HINT_INLINE size_t
 HUF_decodeStreamX2(BYTE* p, BIT_DStream_t* bitDPtr, BYTE* const pEnd,
@@ -1307,7 +1362,7 @@ HUF_decompress1X2_usingDTable_internal_body(
 
     /* decode */
     {   BYTE* const ostart = (BYTE*) dst;
-        BYTE* const oend = ostart + dstSize;
+        BYTE* const oend = ZSTD_maybeNullPtrAdd(ostart, dstSize);
         const void* const dtPtr = DTable+1;   /* force compiler to not use strict-aliasing */
         const HUF_DEltX2* const dt = (const HUF_DEltX2*)dtPtr;
         DTableDesc const dtd = HUF_getDTableDesc(DTable);
@@ -1332,6 +1387,7 @@ HUF_decompress4X2_usingDTable_internal_body(
     const HUF_DTable* DTable)
 {
     if (cSrcSize < 10) return ERROR(corruption_detected);   /* strict minimum : jump table + 1 byte per stream */
+    if (dstSize < 6) return ERROR(corruption_detected);         /* stream 4-split doesn't work */
 
     {   const BYTE* const istart = (const BYTE*) cSrc;
         BYTE* const ostart = (BYTE*) dst;
@@ -1367,7 +1423,7 @@ HUF_decompress4X2_usingDTable_internal_body(
 
         if (length4 > cSrcSize) return ERROR(corruption_detected);  /* overflow */
         if (opStart4 > oend) return ERROR(corruption_detected);     /* overflow */
-        if (dstSize < 6) return ERROR(corruption_detected);         /* stream 4-split doesn't work */
+        assert(dstSize >= 6 /* validated above */);
         CHECK_F( BIT_initDStream(&bitD1, istart1, length1) );
         CHECK_F( BIT_initDStream(&bitD2, istart2, length2) );
         CHECK_F( BIT_initDStream(&bitD3, istart3, length3) );
@@ -1472,11 +1528,11 @@ void HUF_decompress4X2_usingDTable_internal_fast_c_loop(HUF_DecompressFastArgs*
     BYTE* op[4];
     BYTE* oend[4];
     HUF_DEltX2 const* const dtable = (HUF_DEltX2 const*)args->dt;
-    BYTE const* const ilimit = args->ilimit;
+    BYTE const* const ilowest = args->ilowest;
 
     /* Copy the arguments to local registers. */
     ZSTD_memcpy(&bits, &args->bits, sizeof(bits));
-    ZSTD_memcpy(&ip, &args->ip, sizeof(ip));
+    ZSTD_memcpy((void*)(&ip), &args->ip, sizeof(ip));
     ZSTD_memcpy(&op, &args->op, sizeof(op));
 
     oend[0] = op[1];
@@ -1490,13 +1546,12 @@ void HUF_decompress4X2_usingDTable_internal_fast_c_loop(HUF_DecompressFastArgs*
     for (;;) {
         BYTE* olimit;
         int stream;
-        int symbol;
 
         /* Assert loop preconditions */
 #ifndef NDEBUG
         for (stream = 0; stream < 4; ++stream) {
             assert(op[stream] <= oend[stream]);
-            assert(ip[stream] >= ilimit);
+            assert(ip[stream] >= ilowest);
         }
 #endif
         /* Compute olimit */
@@ -1509,7 +1564,7 @@ void HUF_decompress4X2_usingDTable_internal_fast_c_loop(HUF_DecompressFastArgs*
              * We also know that each input pointer is >= ip[0]. So we can run
              * iters loops before running out of input.
              */
-            size_t iters = (size_t)(ip[0] - ilimit) / 7;
+            size_t iters = (size_t)(ip[0] - ilowest) / 7;
             /* Each iteration can produce up to 10 bytes of output per stream.
              * Each output stream my advance at different rates. So take the
              * minimum number of safe iterations among all the output streams.
@@ -1527,8 +1582,8 @@ void HUF_decompress4X2_usingDTable_internal_fast_c_loop(HUF_DecompressFastArgs*
              */
             olimit = op[3] + (iters * 5);
 
-            /* Exit the fast decoding loop if we are too close to the end. */
-            if (op[3] + 10 > olimit)
+            /* Exit the fast decoding loop once we reach the end. */
+            if (op[3] == olimit)
                 break;
 
             /* Exit the decoding loop if any input pointer has crossed the
@@ -1547,59 +1602,63 @@ void HUF_decompress4X2_usingDTable_internal_fast_c_loop(HUF_DecompressFastArgs*
         }
 #endif
 
+#define HUF_4X2_DECODE_SYMBOL(_stream, _decode3)                      \
+    do {                                                              \
+        if ((_decode3) || (_stream) != 3) {                           \
+            int const index = (int)(bits[(_stream)] >> 53);           \
+            HUF_DEltX2 const entry = dtable[index];                   \
+            MEM_write16(op[(_stream)], entry.sequence); \
+            bits[(_stream)] <<= (entry.nbBits) & 0x3F;                \
+            op[(_stream)] += (entry.length);                          \
+        }                                                             \
+    } while (0)
+
+#define HUF_4X2_RELOAD_STREAM(_stream)                                  \
+    do {                                                                \
+        HUF_4X2_DECODE_SYMBOL(3, 1);                                    \
+        {                                                               \
+            int const ctz = ZSTD_countTrailingZeros64(bits[(_stream)]); \
+            int const nbBits = ctz & 7;                                 \
+            int const nbBytes = ctz >> 3;                               \
+            ip[(_stream)] -= nbBytes;                                   \
+            bits[(_stream)] = MEM_read64(ip[(_stream)]) | 1;            \
+            bits[(_stream)] <<= nbBits;                                 \
+        }                                                               \
+    } while (0)
+
+        /* Manually unroll the loop because compilers don't consistently
+         * unroll the inner loops, which destroys performance.
+         */
         do {
-            /* Do 5 table lookups for each of the first 3 streams */
-            for (symbol = 0; symbol < 5; ++symbol) {
-                for (stream = 0; stream < 3; ++stream) {
-                    int const index = (int)(bits[stream] >> 53);
-                    HUF_DEltX2 const entry = dtable[index];
-                    MEM_write16(op[stream], entry.sequence);
-                    bits[stream] <<= (entry.nbBits);
-                    op[stream] += (entry.length);
-                }
-            }
-            /* Do 1 table lookup from the final stream */
-            {
-                int const index = (int)(bits[3] >> 53);
-                HUF_DEltX2 const entry = dtable[index];
-                MEM_write16(op[3], entry.sequence);
-                bits[3] <<= (entry.nbBits);
-                op[3] += (entry.length);
-            }
-            /* Do 4 table lookups from the final stream & reload bitstreams */
-            for (stream = 0; stream < 4; ++stream) {
-                /* Do a table lookup from the final stream.
-                 * This is interleaved with the reloading to reduce register
-                 * pressure. This shouldn't be necessary, but compilers can
-                 * struggle with codegen with high register pressure.
-                 */
-                {
-                    int const index = (int)(bits[3] >> 53);
-                    HUF_DEltX2 const entry = dtable[index];
-                    MEM_write16(op[3], entry.sequence);
-                    bits[3] <<= (entry.nbBits);
-                    op[3] += (entry.length);
-                }
-                /* Reload the bistreams. The final bitstream must be reloaded
-                 * after the 5th symbol was decoded.
-                 */
-                {
-                    int const ctz = ZSTD_countTrailingZeros64(bits[stream]);
-                    int const nbBits = ctz & 7;
-                    int const nbBytes = ctz >> 3;
-                    ip[stream] -= nbBytes;
-                    bits[stream] = MEM_read64(ip[stream]) | 1;
-                    bits[stream] <<= nbBits;
-                }
-            }
+            /* Decode 5 symbols from each of the first 3 streams.
+             * The final stream will be decoded during the reload phase
+             * to reduce register pressure.
+             */
+            HUF_4X_FOR_EACH_STREAM_WITH_VAR(HUF_4X2_DECODE_SYMBOL, 0);
+            HUF_4X_FOR_EACH_STREAM_WITH_VAR(HUF_4X2_DECODE_SYMBOL, 0);
+            HUF_4X_FOR_EACH_STREAM_WITH_VAR(HUF_4X2_DECODE_SYMBOL, 0);
+            HUF_4X_FOR_EACH_STREAM_WITH_VAR(HUF_4X2_DECODE_SYMBOL, 0);
+            HUF_4X_FOR_EACH_STREAM_WITH_VAR(HUF_4X2_DECODE_SYMBOL, 0);
+
+            /* Decode one symbol from the final stream */
+            HUF_4X2_DECODE_SYMBOL(3, 1);
+
+            /* Decode 4 symbols from the final stream & reload bitstreams.
+             * The final stream is reloaded last, meaning that all 5 symbols
+             * are decoded from the final stream before it is reloaded.
+             */
+            HUF_4X_FOR_EACH_STREAM(HUF_4X2_RELOAD_STREAM);
         } while (op[3] < olimit);
     }
 
+#undef HUF_4X2_DECODE_SYMBOL
+#undef HUF_4X2_RELOAD_STREAM
+
 _out:
 
     /* Save the final values of each of the state variables back to args. */
     ZSTD_memcpy(&args->bits, &bits, sizeof(bits));
-    ZSTD_memcpy(&args->ip, &ip, sizeof(ip));
+    ZSTD_memcpy((void*)(&args->ip), &ip, sizeof(ip));
     ZSTD_memcpy(&args->op, &op, sizeof(op));
 }
 
@@ -1611,8 +1670,8 @@ HUF_decompress4X2_usingDTable_internal_fast(
     const HUF_DTable* DTable,
     HUF_DecompressFastLoopFn loopFn) {
     void const* dt = DTable + 1;
-    const BYTE* const iend = (const BYTE*)cSrc + 6;
-    BYTE* const oend = (BYTE*)dst + dstSize;
+    const BYTE* const ilowest = (const BYTE*)cSrc;
+    BYTE* const oend = ZSTD_maybeNullPtrAdd((BYTE*)dst, dstSize);
     HUF_DecompressFastArgs args;
     {
         size_t const ret = HUF_DecompressFastArgs_init(&args, dst, dstSize, cSrc, cSrcSize, DTable);
@@ -1621,16 +1680,19 @@ HUF_decompress4X2_usingDTable_internal_fast(
             return 0;
     }
 
-    assert(args.ip[0] >= args.ilimit);
+    assert(args.ip[0] >= args.ilowest);
     loopFn(&args);
 
     /* note : op4 already verified within main loop */
-    assert(args.ip[0] >= iend);
-    assert(args.ip[1] >= iend);
-    assert(args.ip[2] >= iend);
-    assert(args.ip[3] >= iend);
+    assert(args.ip[0] >= ilowest);
+    assert(args.ip[1] >= ilowest);
+    assert(args.ip[2] >= ilowest);
+    assert(args.ip[3] >= ilowest);
     assert(args.op[3] <= oend);
-    (void)iend;
+
+    assert(ilowest == args.ilowest);
+    assert(ilowest + 6 == args.iend[0]);
+    (void)ilowest;
 
     /* finish bitStreams one by one */
     {
@@ -1679,7 +1741,7 @@ static size_t HUF_decompress4X2_usingDTable_internal(void* dst, size_t dstSize,
     }
 #endif
 
-    if (!(flags & HUF_flags_disableFast)) {
+    if (HUF_ENABLE_FAST_DECODE && !(flags & HUF_flags_disableFast)) {
         size_t const ret = HUF_decompress4X2_usingDTable_internal_fast(dst, dstSize, cSrc, cSrcSize, DTable, loopFn);
         if (ret != 0)
             return ret;

lib/decompress/huf_decompress_amd64.S

@@ -10,11 +10,32 @@
 
 #include "../common/portability_macros.h"
 
+#if defined(__ELF__) && defined(__GNUC__)
 /* Stack marking
  * ref: https://wiki.gentoo.org/wiki/Hardened/GNU_stack_quickstart
  */
-#if defined(__ELF__) && defined(__GNUC__)
 .section .note.GNU-stack,"",%progbits
+
+#if defined(__aarch64__)
+/* Mark that this assembly supports BTI & PAC, because it is empty for aarch64.
+ * See: https://github.com/facebook/zstd/issues/3841
+ * See: https://gcc.godbolt.org/z/sqr5T4ffK
+ * See: https://lore.kernel.org/linux-arm-kernel/20200429211641.9279-8-broonie@kernel.org/
+ * See: https://reviews.llvm.org/D62609
+ */
+.pushsection .note.gnu.property, "a"
+.p2align 3
+.long 4                 /* size of the name - "GNU\0" */
+.long 0x10              /* size of descriptor */
+.long 0x5               /* NT_GNU_PROPERTY_TYPE_0 */
+.asciz "GNU"
+.long 0xc0000000        /* pr_type - GNU_PROPERTY_AARCH64_FEATURE_1_AND */
+.long 4                 /* pr_datasz - 4 bytes */
+.long 3                 /* pr_data - GNU_PROPERTY_AARCH64_FEATURE_1_BTI | GNU_PROPERTY_AARCH64_FEATURE_1_PAC */
+.p2align 3              /* pr_padding - bring everything to 8 byte alignment */
+.popsection
+#endif
+
 #endif
 
 #if ZSTD_ENABLE_ASM_X86_64_BMI2
@@ -131,7 +152,7 @@ HUF_decompress4X1_usingDTable_internal_fast_asm_loop:
     movq 88(%rax), %bits3
     movq 96(%rax), %dtable
     push %rax      /* argument */
-    push 104(%rax) /* ilimit */
+    push 104(%rax) /* ilowest */
     push 112(%rax) /* oend */
     push %olimit   /* olimit space */
 
@@ -156,11 +177,11 @@ HUF_decompress4X1_usingDTable_internal_fast_asm_loop:
     shrq $2, %r15
 
     movq %ip0,     %rax /* rax = ip0 */
-    movq 40(%rsp), %rdx /* rdx = ilimit */
-    subq %rdx,     %rax /* rax = ip0 - ilimit */
-    movq %rax,     %rbx /* rbx = ip0 - ilimit */
+    movq 40(%rsp), %rdx /* rdx = ilowest */
+    subq %rdx,     %rax /* rax = ip0 - ilowest */
+    movq %rax,     %rbx /* rbx = ip0 - ilowest */
 
-    /* rdx = (ip0 - ilimit) / 7 */
+    /* rdx = (ip0 - ilowest) / 7 */
     movabsq $2635249153387078803, %rdx
     mulq %rdx
     subq %rdx, %rbx
@@ -183,9 +204,8 @@ HUF_decompress4X1_usingDTable_internal_fast_asm_loop:
 
     /* If (op3 + 20 > olimit) */
     movq %op3, %rax    /* rax = op3 */
-    addq $20,  %rax    /* rax = op3 + 20 */
-    cmpq %rax, %olimit /* op3 + 20 > olimit */
-    jb .L_4X1_exit
+    cmpq %rax, %olimit /* op3 == olimit */
+    je .L_4X1_exit
 
     /* If (ip1 < ip0) go to exit */
     cmpq %ip0, %ip1
@@ -316,7 +336,7 @@ HUF_decompress4X1_usingDTable_internal_fast_asm_loop:
     /* Restore stack (oend & olimit) */
     pop %rax /* olimit */
     pop %rax /* oend */
-    pop %rax /* ilimit */
+    pop %rax /* ilowest */
     pop %rax /* arg */
 
     /* Save ip / op / bits */
@@ -387,7 +407,7 @@ HUF_decompress4X2_usingDTable_internal_fast_asm_loop:
     movq 96(%rax), %dtable
     push %rax      /* argument */
     push %rax      /* olimit */
-    push 104(%rax) /* ilimit */
+    push 104(%rax) /* ilowest */
 
     movq 112(%rax), %rax
     push %rax /* oend3 */
@@ -414,9 +434,9 @@ HUF_decompress4X2_usingDTable_internal_fast_asm_loop:
 
     /* We can consume up to 7 input bytes each iteration. */
     movq %ip0,     %rax  /* rax = ip0 */
-    movq 40(%rsp), %rdx  /* rdx = ilimit */
-    subq %rdx,     %rax  /* rax = ip0 - ilimit */
-    movq %rax,    %r15   /* r15 = ip0 - ilimit */
+    movq 40(%rsp), %rdx  /* rdx = ilowest */
+    subq %rdx,     %rax  /* rax = ip0 - ilowest */
+    movq %rax,    %r15   /* r15 = ip0 - ilowest */
 
     /* rdx = rax / 7 */
     movabsq $2635249153387078803, %rdx
@@ -426,7 +446,7 @@ HUF_decompress4X2_usingDTable_internal_fast_asm_loop:
     addq %r15, %rdx
     shrq $2, %rdx
 
-    /* r15 = (ip0 - ilimit) / 7 */
+    /* r15 = (ip0 - ilowest) / 7 */
     movq %rdx, %r15
 
     /* r15 = min(r15, min(oend0 - op0, oend1 - op1, oend2 - op2, oend3 - op3) / 10) */
@@ -467,9 +487,8 @@ HUF_decompress4X2_usingDTable_internal_fast_asm_loop:
 
     /* If (op3 + 10 > olimit) */
     movq %op3, %rax    /* rax = op3 */
-    addq $10,  %rax    /* rax = op3 + 10 */
-    cmpq %rax, %olimit /* op3 + 10 > olimit */
-    jb .L_4X2_exit
+    cmpq %rax, %olimit /* op3 == olimit */
+    je .L_4X2_exit
 
     /* If (ip1 < ip0) go to exit */
     cmpq %ip0, %ip1
@@ -537,7 +556,7 @@ HUF_decompress4X2_usingDTable_internal_fast_asm_loop:
     pop %rax /* oend1 */
     pop %rax /* oend2 */
     pop %rax /* oend3 */
-    pop %rax /* ilimit */
+    pop %rax /* ilowest */
     pop %rax /* olimit */
     pop %rax /* arg */
 

lib/decompress/zstd_ddict.c

@@ -14,6 +14,7 @@
 /*-*******************************************************
 *  Dependencies
 *********************************************************/
+#include "../common/allocations.h"  /* ZSTD_customMalloc, ZSTD_customFree */
 #include "../common/zstd_deps.h"   /* ZSTD_memcpy, ZSTD_memmove, ZSTD_memset */
 #include "../common/cpu.h"         /* bmi2 */
 #include "../common/mem.h"         /* low level memory routines */

lib/decompress/zstd_decompress.c

@@ -56,16 +56,18 @@
 *  Dependencies
 *********************************************************/
 #include "../common/zstd_deps.h"   /* ZSTD_memcpy, ZSTD_memmove, ZSTD_memset */
+#include "../common/allocations.h"  /* ZSTD_customMalloc, ZSTD_customCalloc, ZSTD_customFree */
+#include "../common/error_private.h"
+#include "../common/zstd_internal.h"  /* blockProperties_t */
 #include "../common/mem.h"         /* low level memory routines */
+#include "../common/bits.h"  /* ZSTD_highbit32 */
 #define FSE_STATIC_LINKING_ONLY
 #include "../common/fse.h"
 #include "../common/huf.h"
 #include "../common/xxhash.h" /* XXH64_reset, XXH64_update, XXH64_digest, XXH64 */
-#include "../common/zstd_internal.h"  /* blockProperties_t */
 #include "zstd_decompress_internal.h"   /* ZSTD_DCtx */
 #include "zstd_ddict.h"  /* ZSTD_DDictDictContent */
 #include "zstd_decompress_block.h"   /* ZSTD_decompressBlock_internal */
-#include "../common/bits.h"  /* ZSTD_highbit32 */
 
 #if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT>=1)
 #  include "../legacy/zstd_legacy.h"
@@ -244,6 +246,7 @@ static void ZSTD_DCtx_resetParameters(ZSTD_DCtx* dctx)
     dctx->forceIgnoreChecksum = ZSTD_d_validateChecksum;
     dctx->refMultipleDDicts = ZSTD_rmd_refSingleDDict;
     dctx->disableHufAsm = 0;
+    dctx->maxBlockSizeParam = 0;
 }
 
 static void ZSTD_initDCtx_internal(ZSTD_DCtx* dctx)
@@ -264,6 +267,7 @@ static void ZSTD_initDCtx_internal(ZSTD_DCtx* dctx)
 #endif
     dctx->noForwardProgress = 0;
     dctx->oversizedDuration = 0;
+    dctx->isFrameDecompression = 1;
 #if DYNAMIC_BMI2
     dctx->bmi2 = ZSTD_cpuSupportsBmi2();
 #endif
@@ -588,49 +592,52 @@ static size_t readSkippableFrameSize(void const* src, size_t srcSize)
     sizeU32 = MEM_readLE32((BYTE const*)src + ZSTD_FRAMEIDSIZE);
     RETURN_ERROR_IF((U32)(sizeU32 + ZSTD_SKIPPABLEHEADERSIZE) < sizeU32,
                     frameParameter_unsupported, "");
-    {
-        size_t const skippableSize = skippableHeaderSize + sizeU32;
+    {   size_t const skippableSize = skippableHeaderSize + sizeU32;
         RETURN_ERROR_IF(skippableSize > srcSize, srcSize_wrong, "");
         return skippableSize;
     }
 }
 
 /*! ZSTD_readSkippableFrame() :
- * Retrieves a zstd skippable frame containing data given by src, and writes it to dst buffer.
+ * Retrieves content of a skippable frame, and writes it to dst buffer.
  *
  * The parameter magicVariant will receive the magicVariant that was supplied when the frame was written,
  * i.e. magicNumber - ZSTD_MAGIC_SKIPPABLE_START.  This can be NULL if the caller is not interested
  * in the magicVariant.
  *
- * Returns an error if destination buffer is not large enough, or if the frame is not skippable.
+ * Returns an error if destination buffer is not large enough, or if this is not a valid skippable frame.
  *
  * @return : number of bytes written or a ZSTD error.
  */
-ZSTDLIB_API size_t ZSTD_readSkippableFrame(void* dst, size_t dstCapacity, unsigned* magicVariant,
-                                            const void* src, size_t srcSize)
+size_t ZSTD_readSkippableFrame(void* dst, size_t dstCapacity,
+                               unsigned* magicVariant,  /* optional, can be NULL */
+                         const void* src, size_t srcSize)
 {
-    U32 const magicNumber = MEM_readLE32(src);
-    size_t skippableFrameSize = readSkippableFrameSize(src, srcSize);
-    size_t skippableContentSize = skippableFrameSize - ZSTD_SKIPPABLEHEADERSIZE;
+    RETURN_ERROR_IF(srcSize < ZSTD_SKIPPABLEHEADERSIZE, srcSize_wrong, "");
 
-    /* check input validity */
-    RETURN_ERROR_IF(!ZSTD_isSkippableFrame(src, srcSize), frameParameter_unsupported, "");
-    RETURN_ERROR_IF(skippableFrameSize < ZSTD_SKIPPABLEHEADERSIZE || skippableFrameSize > srcSize, srcSize_wrong, "");
-    RETURN_ERROR_IF(skippableContentSize > dstCapacity, dstSize_tooSmall, "");
+    {   U32 const magicNumber = MEM_readLE32(src);
+        size_t skippableFrameSize = readSkippableFrameSize(src, srcSize);
+        size_t skippableContentSize = skippableFrameSize - ZSTD_SKIPPABLEHEADERSIZE;
 
-    /* deliver payload */
-    if (skippableContentSize > 0  && dst != NULL)
-        ZSTD_memcpy(dst, (const BYTE *)src + ZSTD_SKIPPABLEHEADERSIZE, skippableContentSize);
-    if (magicVariant != NULL)
-        *magicVariant = magicNumber - ZSTD_MAGIC_SKIPPABLE_START;
-    return skippableContentSize;
+        /* check input validity */
+        RETURN_ERROR_IF(!ZSTD_isSkippableFrame(src, srcSize), frameParameter_unsupported, "");
+        RETURN_ERROR_IF(skippableFrameSize < ZSTD_SKIPPABLEHEADERSIZE || skippableFrameSize > srcSize, srcSize_wrong, "");
+        RETURN_ERROR_IF(skippableContentSize > dstCapacity, dstSize_tooSmall, "");
+
+        /* deliver payload */
+        if (skippableContentSize > 0  && dst != NULL)
+            ZSTD_memcpy(dst, (const BYTE *)src + ZSTD_SKIPPABLEHEADERSIZE, skippableContentSize);
+        if (magicVariant != NULL)
+            *magicVariant = magicNumber - ZSTD_MAGIC_SKIPPABLE_START;
+        return skippableContentSize;
+    }
 }
 
 /** ZSTD_findDecompressedSize() :
- *  compatible with legacy mode
  *  `srcSize` must be the exact length of some number of ZSTD compressed and/or
  *      skippable frames
- *  @return : decompressed size of the frames contained */
+ *  note: compatible with legacy mode
+ * @return : decompressed size of the frames contained */
 unsigned long long ZSTD_findDecompressedSize(const void* src, size_t srcSize)
 {
     unsigned long long totalDstSize = 0;
@@ -640,9 +647,7 @@ unsigned long long ZSTD_findDecompressedSize(const void* src, size_t srcSize)
 
         if ((magicNumber & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
             size_t const skippableSize = readSkippableFrameSize(src, srcSize);
-            if (ZSTD_isError(skippableSize)) {
-                return ZSTD_CONTENTSIZE_ERROR;
-            }
+            if (ZSTD_isError(skippableSize)) return ZSTD_CONTENTSIZE_ERROR;
             assert(skippableSize <= srcSize);
 
             src = (const BYTE *)src + skippableSize;
@@ -650,17 +655,17 @@ unsigned long long ZSTD_findDecompressedSize(const void* src, size_t srcSize)
             continue;
         }
 
-        {   unsigned long long const ret = ZSTD_getFrameContentSize(src, srcSize);
-            if (ret >= ZSTD_CONTENTSIZE_ERROR) return ret;
+        {   unsigned long long const fcs = ZSTD_getFrameContentSize(src, srcSize);
+            if (fcs >= ZSTD_CONTENTSIZE_ERROR) return fcs;
 
-            /* check for overflow */
-            if (totalDstSize + ret < totalDstSize) return ZSTD_CONTENTSIZE_ERROR;
-            totalDstSize += ret;
+            if (totalDstSize + fcs < totalDstSize)
+                return ZSTD_CONTENTSIZE_ERROR; /* check for overflow */
+            totalDstSize += fcs;
         }
+        /* skip to next frame */
         {   size_t const frameSrcSize = ZSTD_findFrameCompressedSize(src, srcSize);
-            if (ZSTD_isError(frameSrcSize)) {
-                return ZSTD_CONTENTSIZE_ERROR;
-            }
+            if (ZSTD_isError(frameSrcSize)) return ZSTD_CONTENTSIZE_ERROR;
+            assert(frameSrcSize <= srcSize);
 
             src = (const BYTE *)src + frameSrcSize;
             srcSize -= frameSrcSize;
@@ -724,17 +729,17 @@ static ZSTD_frameSizeInfo ZSTD_errorFrameSizeInfo(size_t ret)
     return frameSizeInfo;
 }
 
-static ZSTD_frameSizeInfo ZSTD_findFrameSizeInfo(const void* src, size_t srcSize)
+static ZSTD_frameSizeInfo ZSTD_findFrameSizeInfo(const void* src, size_t srcSize, ZSTD_format_e format)
 {
     ZSTD_frameSizeInfo frameSizeInfo;
     ZSTD_memset(&frameSizeInfo, 0, sizeof(ZSTD_frameSizeInfo));
 
 #if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT >= 1)
-    if (ZSTD_isLegacy(src, srcSize))
+    if (format == ZSTD_f_zstd1 && ZSTD_isLegacy(src, srcSize))
         return ZSTD_findFrameSizeInfoLegacy(src, srcSize);
 #endif
 
-    if ((srcSize >= ZSTD_SKIPPABLEHEADERSIZE)
+    if (format == ZSTD_f_zstd1 && (srcSize >= ZSTD_SKIPPABLEHEADERSIZE)
         && (MEM_readLE32(src) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
         frameSizeInfo.compressedSize = readSkippableFrameSize(src, srcSize);
         assert(ZSTD_isError(frameSizeInfo.compressedSize) ||
@@ -748,7 +753,7 @@ static ZSTD_frameSizeInfo ZSTD_findFrameSizeInfo(const void* src, size_t srcSize
         ZSTD_frameHeader zfh;
 
         /* Extract Frame Header */
-        {   size_t const ret = ZSTD_getFrameHeader(&zfh, src, srcSize);
+        {   size_t const ret = ZSTD_getFrameHeader_advanced(&zfh, src, srcSize, format);
             if (ZSTD_isError(ret))
                 return ZSTD_errorFrameSizeInfo(ret);
             if (ret > 0)
@@ -791,15 +796,17 @@ static ZSTD_frameSizeInfo ZSTD_findFrameSizeInfo(const void* src, size_t srcSize
     }
 }
 
+static size_t ZSTD_findFrameCompressedSize_advanced(const void *src, size_t srcSize, ZSTD_format_e format) {
+    ZSTD_frameSizeInfo const frameSizeInfo = ZSTD_findFrameSizeInfo(src, srcSize, format);
+    return frameSizeInfo.compressedSize;
+}
+
 /** ZSTD_findFrameCompressedSize() :
- *  compatible with legacy mode
- *  `src` must point to the start of a ZSTD frame, ZSTD legacy frame, or skippable frame
- *  `srcSize` must be at least as large as the frame contained
- *  @return : the compressed size of the frame starting at `src` */
+ * See docs in zstd.h
+ * Note: compatible with legacy mode */
 size_t ZSTD_findFrameCompressedSize(const void *src, size_t srcSize)
 {
-    ZSTD_frameSizeInfo const frameSizeInfo = ZSTD_findFrameSizeInfo(src, srcSize);
-    return frameSizeInfo.compressedSize;
+    return ZSTD_findFrameCompressedSize_advanced(src, srcSize, ZSTD_f_zstd1);
 }
 
 /** ZSTD_decompressBound() :
@@ -813,7 +820,7 @@ unsigned long long ZSTD_decompressBound(const void* src, size_t srcSize)
     unsigned long long bound = 0;
     /* Iterate over each frame */
     while (srcSize > 0) {
-        ZSTD_frameSizeInfo const frameSizeInfo = ZSTD_findFrameSizeInfo(src, srcSize);
+        ZSTD_frameSizeInfo const frameSizeInfo = ZSTD_findFrameSizeInfo(src, srcSize, ZSTD_f_zstd1);
         size_t const compressedSize = frameSizeInfo.compressedSize;
         unsigned long long const decompressedBound = frameSizeInfo.decompressedBound;
         if (ZSTD_isError(compressedSize) || decompressedBound == ZSTD_CONTENTSIZE_ERROR)
@@ -833,7 +840,7 @@ size_t ZSTD_decompressionMargin(void const* src, size_t srcSize)
 
     /* Iterate over each frame */
     while (srcSize > 0) {
-        ZSTD_frameSizeInfo const frameSizeInfo = ZSTD_findFrameSizeInfo(src, srcSize);
+        ZSTD_frameSizeInfo const frameSizeInfo = ZSTD_findFrameSizeInfo(src, srcSize, ZSTD_f_zstd1);
         size_t const compressedSize = frameSizeInfo.compressedSize;
         unsigned long long const decompressedBound = frameSizeInfo.decompressedBound;
         ZSTD_frameHeader zfh;
@@ -969,6 +976,10 @@ static size_t ZSTD_decompressFrame(ZSTD_DCtx* dctx,
         ip += frameHeaderSize; remainingSrcSize -= frameHeaderSize;
     }
 
+    /* Shrink the blockSizeMax if enabled */
+    if (dctx->maxBlockSizeParam != 0)
+        dctx->fParams.blockSizeMax = MIN(dctx->fParams.blockSizeMax, (unsigned)dctx->maxBlockSizeParam);
+
     /* Loop on each block */
     while (1) {
         BYTE* oBlockEnd = oend;
@@ -1001,7 +1012,8 @@ static size_t ZSTD_decompressFrame(ZSTD_DCtx* dctx,
         switch(blockProperties.blockType)
         {
         case bt_compressed:
-            decodedSize = ZSTD_decompressBlock_internal(dctx, op, (size_t)(oBlockEnd-op), ip, cBlockSize, /* frame */ 1, not_streaming);
+            assert(dctx->isFrameDecompression == 1);
+            decodedSize = ZSTD_decompressBlock_internal(dctx, op, (size_t)(oBlockEnd-op), ip, cBlockSize, not_streaming);
             break;
         case bt_raw :
             /* Use oend instead of oBlockEnd because this function is safe to overlap. It uses memmove. */
@@ -1014,12 +1026,14 @@ static size_t ZSTD_decompressFrame(ZSTD_DCtx* dctx,
         default:
             RETURN_ERROR(corruption_detected, "invalid block type");
         }
-
-        if (ZSTD_isError(decodedSize)) return decodedSize;
-        if (dctx->validateChecksum)
+        FORWARD_IF_ERROR(decodedSize, "Block decompression failure");
+        DEBUGLOG(5, "Decompressed block of dSize = %u", (unsigned)decodedSize);
+        if (dctx->validateChecksum) {
             XXH64_update(&dctx->xxhState, op, decodedSize);
-        if (decodedSize != 0)
+        }
+        if (decodedSize) /* support dst = NULL,0 */ {
             op += decodedSize;
+        }
         assert(ip != NULL);
         ip += cBlockSize;
         remainingSrcSize -= cBlockSize;
@@ -1049,7 +1063,9 @@ static size_t ZSTD_decompressFrame(ZSTD_DCtx* dctx,
     return (size_t)(op-ostart);
 }
 
-static size_t ZSTD_decompressMultiFrame(ZSTD_DCtx* dctx,
+static
+ZSTD_ALLOW_POINTER_OVERFLOW_ATTR
+size_t ZSTD_decompressMultiFrame(ZSTD_DCtx* dctx,
                                         void* dst, size_t dstCapacity,
                                   const void* src, size_t srcSize,
                                   const void* dict, size_t dictSize,
@@ -1069,7 +1085,7 @@ static size_t ZSTD_decompressMultiFrame(ZSTD_DCtx* dctx,
     while (srcSize >= ZSTD_startingInputLength(dctx->format)) {
 
 #if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT >= 1)
-        if (ZSTD_isLegacy(src, srcSize)) {
+        if (dctx->format == ZSTD_f_zstd1 && ZSTD_isLegacy(src, srcSize)) {
             size_t decodedSize;
             size_t const frameSize = ZSTD_findFrameCompressedSizeLegacy(src, srcSize);
             if (ZSTD_isError(frameSize)) return frameSize;
@@ -1079,6 +1095,15 @@ static size_t ZSTD_decompressMultiFrame(ZSTD_DCtx* dctx,
             decodedSize = ZSTD_decompressLegacy(dst, dstCapacity, src, frameSize, dict, dictSize);
             if (ZSTD_isError(decodedSize)) return decodedSize;
 
+            {
+                unsigned long long const expectedSize = ZSTD_getFrameContentSize(src, srcSize);
+                RETURN_ERROR_IF(expectedSize == ZSTD_CONTENTSIZE_ERROR, corruption_detected, "Corrupted frame header!");
+                if (expectedSize != ZSTD_CONTENTSIZE_UNKNOWN) {
+                    RETURN_ERROR_IF(expectedSize != decodedSize, corruption_detected,
+                        "Frame header size does not match decoded size!");
+                }
+            }
+
             assert(decodedSize <= dstCapacity);
             dst = (BYTE*)dst + decodedSize;
             dstCapacity -= decodedSize;
@@ -1090,17 +1115,18 @@ static size_t ZSTD_decompressMultiFrame(ZSTD_DCtx* dctx,
         }
 #endif
 
-        {   U32 const magicNumber = MEM_readLE32(src);
-            DEBUGLOG(4, "reading magic number %08X (expecting %08X)",
-                        (unsigned)magicNumber, ZSTD_MAGICNUMBER);
+        if (dctx->format == ZSTD_f_zstd1 && srcSize >= 4) {
+            U32 const magicNumber = MEM_readLE32(src);
+            DEBUGLOG(5, "reading magic number %08X", (unsigned)magicNumber);
             if ((magicNumber & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
+                /* skippable frame detected : skip it */
                 size_t const skippableSize = readSkippableFrameSize(src, srcSize);
-                FORWARD_IF_ERROR(skippableSize, "readSkippableFrameSize failed");
+                FORWARD_IF_ERROR(skippableSize, "invalid skippable frame");
                 assert(skippableSize <= srcSize);
 
                 src = (const BYTE *)src + skippableSize;
                 srcSize -= skippableSize;
-                continue;
+                continue; /* check next frame */
         }   }
 
         if (ddict) {
@@ -1316,7 +1342,8 @@ size_t ZSTD_decompressContinue(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, c
             {
             case bt_compressed:
                 DEBUGLOG(5, "ZSTD_decompressContinue: case bt_compressed");
-                rSize = ZSTD_decompressBlock_internal(dctx, dst, dstCapacity, src, srcSize, /* frame */ 1, is_streaming);
+                assert(dctx->isFrameDecompression == 1);
+                rSize = ZSTD_decompressBlock_internal(dctx, dst, dstCapacity, src, srcSize, is_streaming);
                 dctx->expected = 0;  /* Streaming not supported */
                 break;
             case bt_raw :
@@ -1385,6 +1412,7 @@ size_t ZSTD_decompressContinue(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, c
     case ZSTDds_decodeSkippableHeader:
         assert(src != NULL);
         assert(srcSize <= ZSTD_SKIPPABLEHEADERSIZE);
+        assert(dctx->format != ZSTD_f_zstd1_magicless);
         ZSTD_memcpy(dctx->headerBuffer + (ZSTD_SKIPPABLEHEADERSIZE - srcSize), src, srcSize);   /* complete skippable header */
         dctx->expected = MEM_readLE32(dctx->headerBuffer + ZSTD_FRAMEIDSIZE);   /* note : dctx->expected can grow seriously large, beyond local buffer size */
         dctx->stage = ZSTDds_skipFrame;
@@ -1545,6 +1573,7 @@ size_t ZSTD_decompressBegin(ZSTD_DCtx* dctx)
     dctx->litEntropy = dctx->fseEntropy = 0;
     dctx->dictID = 0;
     dctx->bType = bt_reserved;
+    dctx->isFrameDecompression = 1;
     ZSTD_STATIC_ASSERT(sizeof(dctx->entropy.rep) == sizeof(repStartValue));
     ZSTD_memcpy(dctx->entropy.rep, repStartValue, sizeof(repStartValue));  /* initial repcodes */
     dctx->LLTptr = dctx->entropy.LLTable;
@@ -1816,6 +1845,10 @@ ZSTD_bounds ZSTD_dParam_getBounds(ZSTD_dParameter dParam)
             bounds.lowerBound = 0;
             bounds.upperBound = 1;
             return bounds;
+        case ZSTD_d_maxBlockSize:
+            bounds.lowerBound = ZSTD_BLOCKSIZE_MAX_MIN;
+            bounds.upperBound = ZSTD_BLOCKSIZE_MAX;
+            return bounds;
 
         default:;
     }
@@ -1860,6 +1893,9 @@ size_t ZSTD_DCtx_getParameter(ZSTD_DCtx* dctx, ZSTD_dParameter param, int* value
         case ZSTD_d_disableHuffmanAssembly:
             *value = (int)dctx->disableHufAsm;
             return 0;
+        case ZSTD_d_maxBlockSize:
+            *value = dctx->maxBlockSizeParam;
+            return 0;
         default:;
     }
     RETURN_ERROR(parameter_unsupported, "");
@@ -1897,6 +1933,10 @@ size_t ZSTD_DCtx_setParameter(ZSTD_DCtx* dctx, ZSTD_dParameter dParam, int value
             CHECK_DBOUNDS(ZSTD_d_disableHuffmanAssembly, value);
             dctx->disableHufAsm = value != 0;
             return 0;
+        case ZSTD_d_maxBlockSize:
+            if (value != 0) CHECK_DBOUNDS(ZSTD_d_maxBlockSize, value);
+            dctx->maxBlockSizeParam = value;
+            return 0;
         default:;
     }
     RETURN_ERROR(parameter_unsupported, "");
@@ -1908,6 +1948,7 @@ size_t ZSTD_DCtx_reset(ZSTD_DCtx* dctx, ZSTD_ResetDirective reset)
       || (reset == ZSTD_reset_session_and_parameters) ) {
         dctx->streamStage = zdss_init;
         dctx->noForwardProgress = 0;
+        dctx->isFrameDecompression = 1;
     }
     if ( (reset == ZSTD_reset_parameters)
       || (reset == ZSTD_reset_session_and_parameters) ) {
@@ -1924,11 +1965,17 @@ size_t ZSTD_sizeof_DStream(const ZSTD_DStream* dctx)
     return ZSTD_sizeof_DCtx(dctx);
 }
 
-size_t ZSTD_decodingBufferSize_min(unsigned long long windowSize, unsigned long long frameContentSize)
+static size_t ZSTD_decodingBufferSize_internal(unsigned long long windowSize, unsigned long long frameContentSize, size_t blockSizeMax)
 {
-    size_t const blockSize = (size_t) MIN(windowSize, ZSTD_BLOCKSIZE_MAX);
-    /* space is needed to store the litbuffer after the output of a given block without stomping the extDict of a previous run, as well as to cover both windows against wildcopy*/
-    unsigned long long const neededRBSize = windowSize + blockSize + ZSTD_BLOCKSIZE_MAX + (WILDCOPY_OVERLENGTH * 2);
+    size_t const blockSize = MIN((size_t)MIN(windowSize, ZSTD_BLOCKSIZE_MAX), blockSizeMax);
+    /* We need blockSize + WILDCOPY_OVERLENGTH worth of buffer so that if a block
+     * ends at windowSize + WILDCOPY_OVERLENGTH + 1 bytes, we can start writing
+     * the block at the beginning of the output buffer, and maintain a full window.
+     *
+     * We need another blockSize worth of buffer so that we can store split
+     * literals at the end of the block without overwriting the extDict window.
+     */
+    unsigned long long const neededRBSize = windowSize + (blockSize * 2) + (WILDCOPY_OVERLENGTH * 2);
     unsigned long long const neededSize = MIN(frameContentSize, neededRBSize);
     size_t const minRBSize = (size_t) neededSize;
     RETURN_ERROR_IF((unsigned long long)minRBSize != neededSize,
@@ -1936,6 +1983,11 @@ size_t ZSTD_decodingBufferSize_min(unsigned long long windowSize, unsigned long
     return minRBSize;
 }
 
+size_t ZSTD_decodingBufferSize_min(unsigned long long windowSize, unsigned long long frameContentSize)
+{
+    return ZSTD_decodingBufferSize_internal(windowSize, frameContentSize, ZSTD_BLOCKSIZE_MAX);
+}
+
 size_t ZSTD_estimateDStreamSize(size_t windowSize)
 {
     size_t const blockSize = MIN(windowSize, ZSTD_BLOCKSIZE_MAX);
@@ -2131,12 +2183,12 @@ size_t ZSTD_decompressStream(ZSTD_DStream* zds, ZSTD_outBuffer* output, ZSTD_inB
             if (zds->fParams.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN
                 && zds->fParams.frameType != ZSTD_skippableFrame
                 && (U64)(size_t)(oend-op) >= zds->fParams.frameContentSize) {
-                size_t const cSize = ZSTD_findFrameCompressedSize(istart, (size_t)(iend-istart));
+                size_t const cSize = ZSTD_findFrameCompressedSize_advanced(istart, (size_t)(iend-istart), zds->format);
                 if (cSize <= (size_t)(iend-istart)) {
                     /* shortcut : using single-pass mode */
                     size_t const decompressedSize = ZSTD_decompress_usingDDict(zds, op, (size_t)(oend-op), istart, cSize, ZSTD_getDDict(zds));
                     if (ZSTD_isError(decompressedSize)) return decompressedSize;
-                    DEBUGLOG(4, "shortcut to single-pass ZSTD_decompress_usingDDict()")
+                    DEBUGLOG(4, "shortcut to single-pass ZSTD_decompress_usingDDict()");
                     assert(istart != NULL);
                     ip = istart + cSize;
                     op = op ? op + decompressedSize : op; /* can occur if frameContentSize = 0 (empty frame) */
@@ -2158,7 +2210,8 @@ size_t ZSTD_decompressStream(ZSTD_DStream* zds, ZSTD_outBuffer* output, ZSTD_inB
             DEBUGLOG(4, "Consume header");
             FORWARD_IF_ERROR(ZSTD_decompressBegin_usingDDict(zds, ZSTD_getDDict(zds)), "");
 
-            if ((MEM_readLE32(zds->headerBuffer) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {  /* skippable frame */
+            if (zds->format == ZSTD_f_zstd1
+                && (MEM_readLE32(zds->headerBuffer) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {  /* skippable frame */
                 zds->expected = MEM_readLE32(zds->headerBuffer + ZSTD_FRAMEIDSIZE);
                 zds->stage = ZSTDds_skipFrame;
             } else {
@@ -2174,11 +2227,13 @@ size_t ZSTD_decompressStream(ZSTD_DStream* zds, ZSTD_outBuffer* output, ZSTD_inB
             zds->fParams.windowSize = MAX(zds->fParams.windowSize, 1U << ZSTD_WINDOWLOG_ABSOLUTEMIN);
             RETURN_ERROR_IF(zds->fParams.windowSize > zds->maxWindowSize,
                             frameParameter_windowTooLarge, "");
+            if (zds->maxBlockSizeParam != 0)
+                zds->fParams.blockSizeMax = MIN(zds->fParams.blockSizeMax, (unsigned)zds->maxBlockSizeParam);
 
             /* Adapt buffer sizes to frame header instructions */
             {   size_t const neededInBuffSize = MAX(zds->fParams.blockSizeMax, 4 /* frame checksum */);
                 size_t const neededOutBuffSize = zds->outBufferMode == ZSTD_bm_buffered
-                        ? ZSTD_decodingBufferSize_min(zds->fParams.windowSize, zds->fParams.frameContentSize)
+                        ? ZSTD_decodingBufferSize_internal(zds->fParams.windowSize, zds->fParams.frameContentSize, zds->fParams.blockSizeMax)
                         : 0;
 
                 ZSTD_DCtx_updateOversizedDuration(zds, neededInBuffSize, neededOutBuffSize);

lib/decompress/zstd_decompress_block.h

@@ -47,7 +47,7 @@ typedef enum {
  */
 size_t ZSTD_decompressBlock_internal(ZSTD_DCtx* dctx,
                                void* dst, size_t dstCapacity,
-                         const void* src, size_t srcSize, const int frame, const streaming_operation streaming);
+                         const void* src, size_t srcSize, const streaming_operation streaming);
 
 /* ZSTD_buildFSETable() :
  * generate FSE decoding table for one symbol (ll, ml or off)
@@ -64,5 +64,10 @@ void ZSTD_buildFSETable(ZSTD_seqSymbol* dt,
                    unsigned tableLog, void* wksp, size_t wkspSize,
                    int bmi2);
 
+/* Internal definition of ZSTD_decompressBlock() to avoid deprecation warnings. */
+size_t ZSTD_decompressBlock_deprecated(ZSTD_DCtx* dctx,
+                            void* dst, size_t dstCapacity,
+                      const void* src, size_t srcSize);
+
 
 #endif /* ZSTD_DEC_BLOCK_H */

lib/decompress/zstd_decompress_internal.h

@@ -153,6 +153,7 @@ struct ZSTD_DCtx_s
     size_t litSize;
     size_t rleSize;
     size_t staticSize;
+    int isFrameDecompression;
 #if DYNAMIC_BMI2 != 0
     int bmi2;                     /* == 1 if the CPU supports BMI2 and 0 otherwise. CPU support is determined dynamically once per context lifetime. */
 #endif
@@ -166,6 +167,7 @@ struct ZSTD_DCtx_s
     ZSTD_DDictHashSet* ddictSet;                    /* Hash set for multiple ddicts */
     ZSTD_refMultipleDDicts_e refMultipleDDicts;     /* User specified: if == 1, will allow references to multiple DDicts. Default == 0 (disabled) */
     int disableHufAsm;
+    int maxBlockSizeParam;
 
     /* streaming */
     ZSTD_dStreamStage streamStage;

lib/dictBuilder/cover.c

@@ -31,8 +31,8 @@
 #endif
 
 #include "../common/mem.h" /* read */
-#include "../common/pool.h"
-#include "../common/threading.h"
+#include "../common/pool.h" /* POOL_ctx */
+#include "../common/threading.h" /* ZSTD_pthread_mutex_t */
 #include "../common/zstd_internal.h" /* includes zstd.h */
 #include "../common/bits.h" /* ZSTD_highbit32 */
 #include "../zdict.h"
@@ -78,7 +78,7 @@ static clock_t g_time = 0;
 #undef  LOCALDISPLAYUPDATE
 #define LOCALDISPLAYUPDATE(displayLevel, l, ...)                               \
   if (displayLevel >= l) {                                                     \
-    if ((clock() - g_time > g_refreshRate) || (displayLevel >= 4)) {             \
+    if ((clock() - g_time > g_refreshRate) || (displayLevel >= 4)) {           \
       g_time = clock();                                                        \
       DISPLAY(__VA_ARGS__);                                                    \
     }                                                                          \
@@ -301,9 +301,10 @@ static int WIN_CDECL COVER_strict_cmp8(const void *lp, const void *rp) {
  * Returns the first pointer in [first, last) whose element does not compare
  * less than value.  If no such element exists it returns last.
  */
-static const size_t *COVER_lower_bound(const size_t *first, const size_t *last,
+static const size_t *COVER_lower_bound(const size_t* first, const size_t* last,
                                        size_t value) {
-  size_t count = last - first;
+  size_t count = (size_t)(last - first);
+  assert(last >= first);
   while (count != 0) {
     size_t step = count / 2;
     const size_t *ptr = first;
@@ -549,7 +550,8 @@ static void COVER_ctx_destroy(COVER_ctx_t *ctx) {
  */
 static size_t COVER_ctx_init(COVER_ctx_t *ctx, const void *samplesBuffer,
                           const size_t *samplesSizes, unsigned nbSamples,
-                          unsigned d, double splitPoint) {
+                          unsigned d, double splitPoint)
+{
   const BYTE *const samples = (const BYTE *)samplesBuffer;
   const size_t totalSamplesSize = COVER_sum(samplesSizes, nbSamples);
   /* Split samples into testing and training sets */
@@ -733,7 +735,7 @@ static size_t COVER_buildDictionary(const COVER_ctx_t *ctx, U32 *freqs,
   return tail;
 }
 
-ZDICTLIB_API size_t ZDICT_trainFromBuffer_cover(
+ZDICTLIB_STATIC_API size_t ZDICT_trainFromBuffer_cover(
     void *dictBuffer, size_t dictBufferCapacity,
     const void *samplesBuffer, const size_t *samplesSizes, unsigned nbSamples,
     ZDICT_cover_params_t parameters)
@@ -907,8 +909,10 @@ void COVER_best_start(COVER_best_t *best) {
  * Decrements liveJobs and signals any waiting threads if liveJobs == 0.
  * If this dictionary is the best so far save it and its parameters.
  */
-void COVER_best_finish(COVER_best_t *best, ZDICT_cover_params_t parameters,
-                              COVER_dictSelection_t selection) {
+void COVER_best_finish(COVER_best_t* best,
+                      ZDICT_cover_params_t parameters,
+                      COVER_dictSelection_t selection)
+{
   void* dict = selection.dictContent;
   size_t compressedSize = selection.totalCompressedSize;
   size_t dictSize = selection.dictSize;
@@ -980,8 +984,8 @@ COVER_dictSelection_t COVER_selectDict(BYTE* customDictContent, size_t dictBuffe
   size_t largestCompressed = 0;
   BYTE* customDictContentEnd = customDictContent + dictContentSize;
 
-  BYTE * largestDictbuffer = (BYTE *)malloc(dictBufferCapacity);
-  BYTE * candidateDictBuffer = (BYTE *)malloc(dictBufferCapacity);
+  BYTE* largestDictbuffer = (BYTE*)malloc(dictBufferCapacity);
+  BYTE* candidateDictBuffer = (BYTE*)malloc(dictBufferCapacity);
   double regressionTolerance = ((double)params.shrinkDictMaxRegression / 100.0) + 1.00;
 
   if (!largestDictbuffer || !candidateDictBuffer) {
@@ -1119,7 +1123,7 @@ _cleanup:
   free(freqs);
 }
 
-ZDICTLIB_API size_t ZDICT_optimizeTrainFromBuffer_cover(
+ZDICTLIB_STATIC_API size_t ZDICT_optimizeTrainFromBuffer_cover(
     void* dictBuffer, size_t dictBufferCapacity, const void* samplesBuffer,
     const size_t* samplesSizes, unsigned nbSamples,
     ZDICT_cover_params_t* parameters)